No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Quantitative transport mapping (QTM) for differentiating benign and malignant breast lesion: Comparison with traditional kinetics modeling and semi-quantitative enhancement curve characteristics
Abstract
Purpose
To test the feasibility of using quantitative transport mapping (QTM) method, which is based on the inversion of transport equation using spatial deconvolution without any arterial input function, for automatically postprocessing dynamic contrast enhanced MRI (DCE-MRI) to differentiate malignant and benign breast tumors.
Materials and methods
Breast DCE-MRI data with biopsy confirmed malignant (n = 13) and benign tumors (n = 13) was used to assess QTM velocity (|u|) and diffusion coefficient (D), volume transfer constant (Ktrans), volume fraction of extravascular extracellular space (Ve) from kinetics method, and traditional enhancement curve characteristics (ECC: amplitude A, wash-in rate α, wash-out rate β). A Mann-Whitney U test and receiver operating characteristic curve (ROC) analysis were performed to assess the diagnostic performance of these parameters for distinguishing between benign and malignant tumors.
Results
Between malignant and benign tumors, there was a significant difference in |u| and Ktrans, (p = 0.0066, 0.0274, respectively), but not in D, Ve, A, α and β (p = 0.1119, 0.2382, 0.4418,0.2592 and 0.9591, respectively). ROC area-under-the-curve was 0.82, 0.75 (95% confidence level 0.60–0.95, 0.51–0.90) for |u| and Ktrans, respectively.
Conclusion
QTM postprocesses DCE-MRI automatically through deconvolution in space and time to solve the inverse problem of the transport equation. Comparing with traditional kinetics method and ECC, QTM method showed better diagnostic accuracy in differentiating benign from malignant breast tumors in this study.
... 21 Few methods exist for quantifying the spatially varying effective diffusion coefficient of contrast agent in DCE-MRI, 22,23 and similarly few attempt to quantify and investigate interstitial fluid flow (i.e., advection) within the tumor. [24][25][26][27] While there do exist MRI sequences which directly measure interstitial fluid velocity, they struggle with separating vascular flow from interstitial flow and require additional imaging sequences to be applied, significantly increasing the time a patient spends in the scanner. 28,29 Given that DCE-MRI is becoming a standard clinical practice and novel methods for its acquisition are being actively developed, 30 there is growing opportunity to utilize it for studying interstitial flow, without the need for requiring more time in the scanner. ...
... 33,34 Weak-form methods bypass the use of discrete approximations of derivatives on noisy or sparse data, which are used by the original SINDy implementation 31 or gradient descent methods used in Jacobian estimation for standard ODE-fitting techniques. [24][25][26] The key insight from weak-form methods is the integration of raw data with known basis functions and their derivatives, selected for the problem at hand. 33,34 However, existing weak-form methods recover global partial differential equation (PDE) coefficients and do not recover spatially varying parameter fields, including IFF. ...
... Â 10 À3 6 7.04 Â 10 À4 mm/s) with results from recent studies that used phase-contrast imaging to directly measure fluid velocity within tumors (1.10 Â 10 À1 6 5.5 Â 10 À4 and 1.10 Â 10 À1 6 5.5 Â 10 À4 mm/s), 28,29 we find some discrepancy, likely due to contribution from vascular flow, which could not be disambiguated from tissue interstitial fluid flow in the phase-contrast methods resulting in higher values. 24,25,36,37 In the present study, the bolus arrival time is corrected for in each voxel, thus minimizing the contribution of vascular velocity in the total velocity field. Additionally, we demonstrate (e) Distribution of the in-plane velocity of the entire enhancing glioblastoma and resection cavity (mean ¼ 5.23 Â 10 À1 6 5.10 Â 10 À1 ). ...
Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a routine method to noninvasively quantify perfusion dynamics in tissues. The standard practice for analyzing DCE-MRI data is to fit an ordinary differential equation to each voxel. Recent advances in data science provide an opportunity to move beyond existing methods to obtain more accurate measurements of fluid properties. Here, we developed a localized convolutional function regression that enables simultaneous measurement of interstitial fluid velocity, diffusion, and perfusion in 3D. We validated the method computationally and experimentally, demonstrating accurate measurement of fluid dynamics in situ and in vivo. Applying the method to human MRIs, we observed tissue-specific differences in fluid dynamics, with an increased fluid velocity in breast cancer as compared to brain cancer. Overall, our method represents an improved strategy for studying interstitial flows and interstitial transport in tumors and patients. We expect that our method will contribute to the better understanding of cancer progression and therapeutic response.
... In theory, this bias can be removed by the use of spatiotemporal PK models (Sourbron 2014). Implementations of this approach have mainly focused on one-compartment models with transport by diffusion (Koh 2013), convection (Zhou et al 2021, Zhang et al 2023, or both (Sourbron 2015, Elkin et al 2019, Zhang et al 2022. Hybrid approaches have also been proposed, coupling a one-compartment spatiotemporal model for interstitial transport with vascular delivery modeled by a single, global AIF (Pellerin et al 2007, Fluckiger et al 2013, Sinno et al 2021, Sainz-DeMena et al 2022, Sinno et al 2022. ...
... All the above implementations apply additional constraints on the reconstructed model parameters (Shalom et al 2023(Shalom et al , 2024a(Shalom et al , 2024b, for instance an assumption that diffusion is constant in space (Pellerin et al 2007), that the diffusion gradient between adjacent voxels is negligible (Fluckiger et al 2013), that parameter fields have small spatial gradients (Liu et al 2021, Zhou et al 2021, Zhang et al 2022, 2023, that transport is only radial in a lesion (Sinno et al 2021(Sinno et al , 2022, that perfusion is modeled by Darcy flow (Naevdal et al 2016), or that parameter fields are in a known relationship to each other (Naevdal et al 2016). Constraints of this type are included to reduce the computational complexity, but it is not always clear that they are physically justified, creating a risk of new biases. ...
... The solid black line indicates the ground truth, and the colored dots show reconstructions with different initial guesses. (Liu et al 2021, Zhou et al 2021, Zhang et al 2022, 2023, fixing less critical parameters to literature values (Pellerin et al 2007), or reverting to a measured AIF at the boundaries of the imaging slab. The use of physical constraints derived from principles of fluid dynamics and porous media theory presents a particularly attractive approach as it also provides a mechanism for studying the mechanical properties of physiological flow. ...
Objective. Standard models for perfusion quantification in DCE-MRI produce a bias by treating voxels as isolated systems. Spatiotemporal models can remove this bias, but it is unknown whether they are fundamentally identifiable. The aim of this study is to investigate this question in silico using one-dimensional toy systems with a one-compartment blood flow model and a two-compartment perfusion model. Approach. For each of the two models, identifiability is explored theoretically and in-silico for three systems. Concentrations over space and time are simulated by forward propagation. Different levels of noise and temporal undersampling are added to investigate sensitivity to measurement error. Model parameters are fitted using a standard gradient descent algorithm, applied iteratively with a stepwise increasing time window. Model fitting is repeated with different initial values to probe uniqueness of the solution. Reconstruction accuracy is quantified for each parameter by comparison to the ground truth. Main results. Theoretical analysis shows that flows and volume fractions are only identifiable up to a constant, and that this degeneracy can be removed by proper choice of parameters. Simulations show that in all cases, the tissue concentrations can be reconstructed accurately. The one-compartment model shows accurate reconstruction of blood velocities and arterial input functions, independent of the initial values and robust to measurement error. The two-compartmental perfusion model was not fully identifiable, showing good reconstruction of arterial velocities and input functions, but multiple valid solutions for the perfusion parameters and venous velocities, and a strong sensitivity to measurement error in these parameters. Significance. These results support the use of one-compartment spatiotemporal flow models, but two-compartment perfusion models were not sufficiently identifiable. Future studies should investigate whether this degeneracy is resolved in more realistic 2D and 3D systems, by adding physically justified constraints, or by optimizing experimental parameters such as injection duration or temporal resolution.
... 13 To address this problem, we proposed to model changes in spatiotemporal tracer concentration according to the mass transport equation that utilizes spatial and temporal derivatives of the concentration without the selection of an AIF. 12 Blood flow velocity can be calculated fully automatedly from fitting four dimensional (4D) dynamic tracer imaging data to the transport equation, which is termed as quantitative transport mapping (QTM). 12 It has been demonstrated that 1) QTM velocity is more accurate than traditional Kety flow for blood flow quantification in in silico validation; 12,14 2) QTM velocity has a significant value in identifying breast cancer malignancy, 15 nasopharyngeal cancer gene expressions 16 , lung shunt fraction 17 , and progressive liver disease stages. 14 Given its promising diagnostic value in various diseases, we applied this technique to AD in this work for the first time to evaluate its ability of early detection. ...
... The quantitative transport mapping was modeled by the mass conservation equation of tracer 12,14,15 : ...
... For perfusion estimation, ( ) could be considered negligible since diffusion effects are at much slower rate than blood perfusion. The reconstruction of perfusion velocity is then performed following the optimization below 12,14,15 : ...
Background
Quantitative transport mapping (QTM) of blood velocity, based on the transport equation has been demonstrated higher accuracy and sensitivity of perfusion quantification than the traditional Kety's method-based blood flow (Kety flow). This study aimed to investigate the associations between QTM velocity and cognitive function in Alzheimer's disease (AD) using multiple post-labeling delay arterial spin labeling (ASL) MRI.
Methods
A total of 128 subjects (21 normal controls (NC), 80 patients with mild cognitive impairment (MCI), and 27 AD) were recruited prospectively. All participants underwent MRI examination and neuropsychological evaluation. QTM velocity and traditional Kety flow maps were computed from multiple delay ASL. Regional quantitative perfusion measurements were performed and compared to study group differences. We tested the hypothesis that cognition declines with reduced cerebral blood flow with consideration of age and gender effects.
Results
In cortical gray matter (GM) and the hippocampus, QTM velocity and Kety flow showed decreased values in AD group compared to NC and MCI groups; QTM velocity, but not Kety flow, showed a significant difference between MCI and NC groups. QTM velocity and Kety flow showed values decreasing with age; QTM velocity, but not Kety flow, showed a significant gender difference between male and female. QTM velocity and Kety flow in the hippocampus were positively correlated with cognition, including global cognition, memory, executive function, and language function.
Conclusion
This study demonstrated an increased sensitivity of QTM velocity as compared with the traditional Kety flow. Specifically, we observed only in QTM velocity, reduced perfusion velocity in GM and the hippocampus in MCI compared with NC. Both QTM velocity and Kety flow demonstrated reduction in AD vs controls. Decreased QTM velocity and Kety flow in the hippocampus were
correlated with cognitive measures. These findings suggest QTM velocity as an improved biomarker for early AD blood flow alterations.
... Zhang et al., 51 Sourbron 53 and Liu et al. 54-56 ...
... The Cornell group, applying a similar inverse approach, 50,52 developed their method to include diffusive transport. 51 Experiments include 3D clinical breast DCE-MRI data, where Zhang et al. 51 reported a more statistically significant distinction between malignant and benign breast tumors in u p than K trans from the Tofts model. Sourbron 53 introduced a one-compartment model with both vascular convection and diffusion ( Figure 2F). ...
... The Cornell group, applying a similar inverse approach, 50,52 developed their method to include diffusive transport. 51 Experiments include 3D clinical breast DCE-MRI data, where Zhang et al. 51 reported a more statistically significant distinction between malignant and benign breast tumors in u p than K trans from the Tofts model. Sourbron 53 introduced a one-compartment model with both vascular convection and diffusion ( Figure 2F). ...
In perfusion MRI, image voxels form a spatially organized network of systems, all exchanging indicator with their immediate neighbors. Yet the current paradigm for perfusion MRI analysis treats all voxels or regions‐of‐interest as isolated systems supplied by a single global source. This simplification not only leads to long‐recognized systematic errors but also fails to leverage the embedded spatial structure within the data. Since the early 2000s, a variety of models and implementations have been proposed to analyze systems with between‐voxel interactions. In general, this leads to large and connected numerical inverse problems that are intractible with conventional computational methods. With recent advances in machine learning, however, these approaches are becoming practically feasible, opening up the way for a paradigm shift in the approach to perfusion MRI. This paper seeks to review the work in spatiotemporal modelling of perfusion MRI using a coherent, harmonized nomenclature and notation, with clear physical definitions and assumptions. The aim is to introduce clarity in the state‐of‐the‐art of this promising new approach to perfusion MRI, and help to identify gaps of knowledge and priorities for future research.
... Recently, the quantitative transport mapping (QTM) method [14,15] has been developed to estimate mass flux characterized by velocity |u| without a global arterial input function (AIF) used in traditional Kety's tracer kinetic analysis [16,17]. QTM velocity has been shown to be more accurate than Kety's parameters in validation with numerical ground truth [14] and more sensitive than Kety's parameters for differentiating benign from malignant tumors compared with biopsy [18,19]. Accordingly, we propose to investigate the feasibility of noninvasive prediction of LSF according to QTM velocity |u|, as well as Kety's parameters, derived from DCE MRI. ...
... In the patient cohort, QTM velocity |u| but not Kety's parameters demonstrated significant correlation with LSF. Although Kety's method has not been applied to lung shunting fraction estimation in previous studies, this result is consistent with the hypothesis that increased artery-vein connections bypassing capillaries (shunts) increases the mean liver blood velocity, and also consistent with previous reports showing QTM improves upon Kety's method by replacing a global arterial input function in Kety's model with the local mass flux gradient in QTM [14,18,19]. In addition to a significantly higher velocity |u| observed in high LSF group, we also observed an increase in Kety's parameters K trans and V e in high LSF group, which may reflect a higher tissue exchange rate and EES space in tumors with abnormal vasculature compared to normal tissue [34,35]. ...
... For example, texture analysis or radiomic features of the transport quantity maps, deep learning and larger data sets may help construct the LSF prediction model. We combined QTM velocity ROI values with Kety's parameter ROI values but the combination failed to improve the LSF prediction accuracy, which may suggest that QTM already contains all the information in Kety's parameters as well as better information than Kety's parameters, consistent with previous publications [14,18,19]. Historically, Tc-99m-MAA has been a well-developed radiotracer for pulmonary vascular imaging, especially for studying pulmonary embolism since 1960s [43] and has been conveniently adopted for estimating LSF since the beginning of radioembolization practice [44][45][46]. ...
There is no noninvasive method to estimate lung shunting fraction (LSF) in patients with liver tumors undergoing Yttrium-90 (Y90) therapy. We propose to predict LSF from noninvasive dynamic contrast enhanced (DCE) MRI using perfusion quantification. Two perfusion quantification methods were used to process DCE MRI in 25 liver tumor patients: Kety’s tracer kinetic modeling with a delay-fitted global arterial input function (AIF) and quantitative transport mapping (QTM) based on the inversion of transport equation using spatial deconvolution without AIF. LSF was measured on SPECT following Tc-99m macroaggregated albumin (MAA) administration via hepatic arterial catheter. The patient cohort was partitioned into a low-risk group (LSF ≤ 10%) and a high-risk group (LSF > 10%). Results: In this patient cohort, LSF was positively correlated with QTM velocity |u| (r = 0.61, F = 14.0363, p = 0.0021), and no significant correlation was observed with Kety’s parameters, tumor volume, patient age and gender. Between the low LSF and high LSF groups, there was a significant difference for QTM |u| (0.0760 ± 0.0440 vs. 0.1822 ± 0.1225 mm/s, p = 0.0011), and Kety’s Ktrans (0.0401 ± 0.0360 vs 0.1198 ± 0.3048, p = 0.0471) and Ve (0.0900 ± 0.0307 vs. 0.1495 ± 0.0485, p = 0.0114). The area under the curve (AUC) for distinguishing between low LSF and high LSF was 0.87 for |u|, 0.80 for Ve and 0.74 for Ktrans. Noninvasive prediction of LSF is feasible from DCE MRI with QTM velocity postprocessing.
... Few methods exist for quantifying the spatially varying effective diffusion coefficient of contrast agent in DCE-MRI (19,20), and even fewer attempt to quantify and investigate interstitial fluid flow (i.e. advection) within the tumor (21)(22)(23). While there exist MR sequences which attempt to measure interstitial fluid velocity (24,25), they largely remain research tools. ...
... Basis functions are selected such that they possess analytical expressions for derivatives, and can be constructed from polynomials (30). This method bypasses the use of discrete approximations of derivatives on noisy or sparse data, which are used by SINDy (28), or iterative gradient descent methods used in modern DCE-MRI model inversion which require estimating derivatives of the objective function with respect to the data (21)(22)(23). ...
Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a routine method to non-invasively quantify perfusion dynamics in tissues. The standard practice for analyzing DCE-MRI data is to fit an ordinary differential equation to each voxel. Recent advances in data science provide an opportunity to move beyond existing methods to obtain more accurate measurements of fluid properties. Here we present localized convolutional function regression, simultaneously measuring interstitial fluid velocity, diffusion, and perfusion in 3D. We validate the method computationally and experimentally, demonstrating accurate measurement of fluid dynamics in situ and in vivo. Applying the method to human MRIs, we observe tissue-specific differences in fluid dynamics, with an increased fluid velocity in breast cancer as compared to brain cancer.
One-Sentence Summary
A physics-informed computational method enables accurate and efficient measurement of fluid dynamics in individual patient tumors and demonstrates differences between tissues.
... Using a vascular tree CFD simulation to validate quantitative tissue perfusion, QTM is shown to be substantially more accurate than Kety's approach for kidney perfusion quantification [15]. Correlating with immunohistochemistry, QTM processing of DCE-MRI provides more significant resolutions of pathological markers than Kety's approach in nasopharyngeal carcinoma [18] and in breast cancer [19]. ...
... This suggests that the fluid mechanics approach to tissue perfusion may be scaled to any organ tissue. There have been CFD studies for liver perfusion and drug delivery (16)(17)(18) using microvascular network (15,17,19), which can serve as a ground truth for validating tissue perfusion quantification (15,20) through integration of microvasculature over a voxel for interpreting the contrast agent concentration changes in space and time (21). However, these studies use plug flow for vascular branches, possibly due to limitations in memory and computational power [28], [56]. ...
Objective:
We quantify liver perfusion using quantitative transport mapping (QTM) method that is free of arterial input function (AIF). QTM method is validated in a vasculature computational fluid dynamics (CFD) simulation and is applied for processing dynamic contrast enhanced (DCE) MRI images in differentiating liver with nonalcoholic fatty liver disease (NAFLD) from healthy controls using pathology reference in a preclinical rabbit model.
Methods:
QTM method was validated on a liver perfusion simulation based on fluid dynamics using a rat liver vasculature model and the mass transport equation. In the NAFLD grading task, DCE MRI images of 7 adult rabbits with methionine choline-deficient diet-induced nonalcoholic steatohepatitis (NASH), 8 adult rabbits with simple steatosis (SS) were acquired and processed using QTM method and dual-input two compartment Kety's method respectively. Statistical analysis was performed on six perfusion parameters: velocity magnitude [Formula: see text] derived from QTM, liver arterial blood flow [Formula: see text], liver venous blood flow [Formula: see text], permeability [Formula: see text], blood volume [Formula: see text] and extravascular space volume [Formula: see text] averaged in liver ROI.
Results:
In the simulation, QTM method successfully reconstructed blood flow, reduced error by 48% compared to Kety's method. In the preclinical study, only QTM |u| showed significant difference between high grade NAFLD group and low grade NAFLD group.
Conclusion:
QTM postprocesses DCE-MRI automatically through deconvolution in space and time to solve the inverse problem of the transport equation. Comparing with Kety's method, QTM method showed higher accuracy and better differentiation in NAFLD classification task.
Significance:
We propose to apply QTM method in liver DCE MRI perfusion quantification.
... Koundal et al. [30] use optimal mass transport combined with a spatially constant diffusion. Optimization approaches were proposed [37,38,66] to estimate the advection-diffusion parameters of an advectiondiffusion equation in 3D. Though promising, the numerical optimization approach is time-consuming, especially when dealing with large datasets. ...
... However, Zhou et al. [67,68] assume the diffusion process is negligible; therefore, only the velocity field is estimated similar to optical flow [8,26,50]. Liu et al. [37,38] and Zhang et al. [66] estimate both the spatial-varying ve-locity and diffusion fields, yet modeling the diffusion as a scalar field, which cannot express the important diffusion anisotropy. ...
Advection-diffusion equations describe a large family of natural transport processes, e.g., fluid flow, heat transfer, and wind transport. They are also used for optical flow and perfusion imaging computations. We develop a machine learning model, D^2-SONATA, built upon a stochastic advection-diffusion equation, which predicts the velocity and diffusion fields that drive 2D/3D image time-series of transport. In particular, our proposed model incorporates a model of transport atypicality, which isolates abnormal differences between expected normal transport behavior and the observed transport. In a medical context such a normal-abnormal decomposition can be used, for example, to quantify pathologies. Specifically, our model identifies the advection and diffusion contributions from the transport time-series and simultaneously predicts an anomaly value field to provide a decomposition into normal and abnormal advection and diffusion behavior. To achieve improved estimation performance for the velocity and diffusion-tensor fields underlying the advection-diffusion process and for the estimation of the anomaly fields, we create a 2D/3D anomaly-encoded advection-diffusion simulator, which allows for supervised learning. We further apply our model on a brain perfusion dataset from ischemic stroke patients via transfer learning. Extensive comparisons demonstrate that our model successfully distinguishes stroke lesions (abnormal) from normal brain regions, while reconstructing the underlying velocity and diffusion tensor fields.
... Fusco et al. reported that malignant had more blood vessels and deoxyhemoglobin than benign [12]. Zhang et al. reported that the speed of benign Quantitative Transport Mapping (QTM) was lower than that of malignant [13]. While Fuller et al. reported that vascular density was significantly reduced in high-grade tumors [14]. ...
Objective: Evaluating the diagnostic performance of SVM to classify benign and malignant by performing a meta-analysis. Methods:The data used for this study were secondary data. It consisted of 221 mammogram images (mean age 57.5 years) with 164 malignant and 57 benign, taken from a radiological database that has been examined by a radiologist with more than 20 years of experience.Also, histopathological record data that had beenexamined by an oncologist with more than 20 years of experience. Mammograms were taken from January 2022 to June 2022.In all, 221 mammograms consisting of 164 malignant and 57 benign were used as SVM method training, and 20 mammograms consisting of 10 malignant and 10 benign were used to test the performance of the SVM method. It was then evaluated using pathology results as the gold standard
Results: Benign had a significantly lower deviation(an averageof 29.2661230 ± 10.14916673) than malignant (an average of 33.1841234 ± 11.70238757). The SVM method performance value obtained the values of TP, FP, TN, FN, accuracy, sensitivity, Specificity, and Precision, respectively 7,7, 3, 3, 50%, 70%, 30%, and 50%.Conclusion: A proper performance to distinguish benign and malignant can be obtained using the physical deviation parameters with the SVM classification approach. However, these findings should be proven in larger datasets with different mammographic scanners. Our meta-analysis shows that the physical parameters and SVM have high sensitivity but low specificity. Of the nine physical parameters in the mammogram, only the parameter deviation was significant to distinguish between benign and malignant. TheSVM method proved to be able to differentiate between benign and malignant
... Fusco et al. reported that malignant had more blood vessels and deoxyhemoglobin than benign [12]. Zhang et al. reported that the speed of benign Quantitative Transport Mapping (QTM) was lower than that of malignant [13]. While Fuller et al. reported that vascular density was significantly reduced in high-grade tumors [14]. ...
Objective: Evaluating the diagnostic performance of SVM to classify benign and malignant by
performing a meta-analysis.
Methods: The data used for this study were secondary data. It consisted of 221 mammogram
images (mean age 57.5 years) with 164 malignant and 57 benign, taken from a radiological
database that has been examined by a radiologist with more than 20 years of experience. Also,
histopathological record data that had been examined by an oncologist with more than 20 years of
experience. Mammograms were taken from January 2022 to June 2022. In all, 221 mammograms
consisting of 164 malignant and 57 benign were used as SVM method training, and 20
mammograms consisting of 10 malignant and 10 benign were used to test the performance of the
SVM method. It was then evaluated using pathology results as the gold standard.
Results: Benign had a significantly lower deviation (an average of 29.2661230 ± 10.14916673)
than malignant (an average of 33.1841234 ± 11.70238757). The SVM method performance value
obtained the values of TP, FP, TN, FN, accuracy, sensitivity, Specificity, and Precision, respectively
7,7, 3, 3, 50%, 70%, 30%, and 50%.
Conclusion: A proper performance to distinguish benign and malignant can be obtained using the
physical deviation parameters with the SVM classification approach. However, these findings
should be proven in larger datasets with different mammographic scanners. Our meta-analysis
shows that the physical parameters and SVM have high sensitivity but low specificity. Of the nine
physical parameters in the mammogram, only the parameter deviation was significant to distinguish
between benign and malignant. The SVM method proved to be able to differentiate between benign
and malignant.
Breast cancer is a commonly diagnosed disease in women. This research aimed to create a transformed image features can improve machine learning performance for detecting benign-malignant of breast cancer.This research was quantitative research. Data was taken from the radiology installation at Doctor Sutomo Hospital from 2010 until now, where there were 670 data, consisting of 342 benign and 328 malignant. The data was distributed randomly; 70% was used for training, while the remaining 30% was used for testing. Every mammography had nine features: Entropy, Entropy of hdiff, contrast, Angular second moment, Angular second moment of hdiff Inverse difference moment, mean, Mean hdiff, and deviation. This research developed each feature into ten sub-features, namely Entropy at a distance of 1 pixel to Entropy at a distance of 10 pixels, and so on until the mean Hdiff at a distance of 10 pixels. Thus, the total features used in this research were 90 features. This research used three types of transformation, namely original, binary transformation, and bipolar transformation. Besides, this research also used three types of methods, namely 90 features, average, and optimization. Furthermore, this research sought the best performance based on the widest ROC graph, highest accuracy, and lowest false negative rate. In addition, this research also sought the best types of transformation and methods. Models with optimization types with binary and bipolar transformations had the highest positive true values. Models with optimization types with binary and bipolar transformations both had the lowest false negative values. The average type model with bipolar transformation had the highest accuracy value, followed by binary transformation. Models with optimization types with binary and bipolar transformations both had the highest ROC area. Based on the three methods and three transformations proposed, it was found that the optimization method and types of binary and bipolar transformations had the best performance.
Breast cancer's impact necessitates refined diagnostic approaches. This study develops a nomogram using radiology quantitative features from contrast-enhanced cone-beam breast CT for accurate preoperative classification of benign and malignant breast tumors.
A retrospective study enrolled 234 females with breast tumors, split into training and test sets. Contrast-enhanced cone-beam breast CT-images were acquired using Koning Breast CT-1000. Quantitative assessment features were extracted via 3D-slicer software, identifying independent predictors. The nomogram was constructed to preoperative differentiation benign and malignant breast tumors. Calibration curve was used to assess whether the model showed favorable correspondence with pathological confirmation. Decision curve analysis confirmed the model's superiority.
The study enrolled 234 female patients with a mean age of 50.2 years (SD ± 9.2). The training set had 164 patients (89 benign, 75 malignant), and the test set had 70 patients (29 benign, 41 malignant). The nomogram achieved excellent predictive performance in distinguishing benign and malignant breast lesions with an AUC of 0.940 (95% CI 0.900–0.940) in the training set and 0.970 (95% CI 0.940–0.970) in the test set.
This study illustrates the effectiveness of quantitative radiology features derived from contrast-enhanced cone-beam breast CT in distinguishing between benign and malignant breast tumors. Incorporating these features into a nomogram-based diagnostic model allows for breast tumor diagnoses that are objective and possess good accuracy. The application of these insights could substantially increase reliability and efficacy in the management of breast tumors, offering enhanced diagnostic capability.
Digital mammography is a tool for early detection of breast cancer, but its sensitivity is low in dense breasts, and the false negatives are high. Therefore, this research aims to create a digital mammography data transformation model to improve the accuracy of benign and malignant detection in breast and compare it with other advanced methods. This research took data from Dokter Sutomo Hospital Surabaya and Sanglah Hospital Denpasar in the form of mammogram images taken from the hospital database with complete pathology results from 2010 to 2023. A total of 442 mammograms consisting of 114 benign and 328 malignant, 50% taken for training as many as 57 benign and 164 malignant, and 50% for trials as many as 57 benign and 164 malignant. This research used three methods: K-Nearest Neighbor (KNN), Naïve Bayes, and Support Vector Machine (SVM). Then, this research compared the three methods. Inclusion criteria: complete pathology results, no radiotherapy, and chemotherapy. The local hospital ethics committee approved this research. All patients were informed and obtained their verbal consent.The proposed KNN method with binary transformation had the best value with a sensitivity of 96.66% and False Negative 0%, compared to the Naïve Bayes method, which had a sensitivity value of 94.44% and False Negative 4%. In comparison, the SVM method had a sensitivity value of 84.85% and False Negative 4.16 %. Our meta-analysis showed that transformed physical parameters could increase sensitivity and decrease False Negatives. However, these findings must be proven on larger and multiple datasets with different mammography scanners
Digital mammography is a tool for early detection of breast cancer, but its sensitivity is low in dense breasts, and the false negatives are high. Therefore, this research aims to create a digital mammography data transformation model to improve the accuracy of benign and malignant detection in breast and compare it with other advanced methods. This research took data from Dokter Sutomo Hospital Surabaya and Sanglah Hospital Denpasar in the form of mammogram images taken from the hospital database with complete pathology results from 2010 to 2023. A total of 442 mammograms consisting of 114 benign and 328 malignant, 50% taken for training as many as 57 benign and 164 malignant, and 50% for trials as many as 57 benign and 164 malignant. This research used three methods: K-Nearest Neighbor (KNN), Naïve Bayes, and Support Vector Machine (SVM). Then, this research compared the three methods. Inclusion criteria: complete pathology results, no radiotherapy, and chemotherapy. The local hospital ethics committee approved this research. All patients were informed and obtained their verbal consent.The proposed KNN method with binary transformation had the best value with a sensitivity of 96.66% and False Negative 0%, compared to the Naïve Bayes method, which had a sensitivity value of 94.44% and False Negative 4%. In comparison, the SVM method had a sensitivity value of 84.85% and False Negative 4.16 %. Our meta-analysis showed that transformed physical parameters could increase sensitivity and decrease False Negatives. However, these findings must be proven on larger and multiple datasets with different mammography scanners
Recently a new family of loss functions called smart error sums has been suggested. These loss functions account for correlations within experimental data and force modeled data to obey these correlations. As a result, multiplicative systematic errors of experimental data can be revealed and corrected. The smart error sums are based on 2D correlation analysis which is a comparably recent methodology for analyzing spectroscopic data that has found broad application. In this contribution we mathematically generalize and break down this methodology and the smart error sums to uncover the mathematic roots and simplify it to craft a general tool beyond spectroscopic modelling. This reduction also allows a simplified discussion about limits and prospects of this new method including one of its potential future uses as a sophisticated loss function in deep learning. To support its deployment, the work includes computer code to allow reproduction of the basic results.
To achieve better treatment regimen and follow-up assessment design for intensity-modulated radiotherapy (IMRT) treated nasopharyngeal carcinoma (NPC) patients, an accurate progression free survival (PFS) time prediction algorithm is needed. We propose to develop PFS prediction model of NPC patients after IMRT treatment using deep learning method, and to compare that with traditional texture analysis method. 151 NPC patients were included in this retrospective study. T1 weighted, proton density and dynamic contrast enhanced MR images were acquired. Expression level of five genes (HIF-1α, EGFR, PTEN, Ki-67, VEGF) and infection of Epstein-Barr virus were tested. A residual network was trained to predict PFS from magnetic resonance (MR) images. The output as well as patient characteristics were combined using linear regression model to give a final PFS prediction. The prediction accuracy was compared with traditional texture analysis method. Regression model combining deep learning output with HIF-1α expression and EB infection gives the best PFS prediction accuracy (Spearman correlation R2 =0.53; Harrell's C-index = 0.82; ROC analysis AUC=0.88; log rank test HR= 8.45), higher than regression model combining texture analysis with HIF-1α expression (Spearman correlation R2 =0.14; Harrell's C-index =0.68; ROC analysis AUC=0.76; log rank test HR= 2.85). Deep learning method doesn't require manually drawn tumor ROI, thus is fully automatic. MR image processing using deep learning combined with patient characteristics can provide accurate PFS prediction for nasopharyngeal carcinoma patients and doesn't rely on specific kernels or tumor ROI as texture analysis method.
Objective: Evaluating the diagnostic performance of SVM to classify benign and malignant by performing a meta-analysis. Methods: The data used for this study were secondary data. It consisted of 221 mammogram images (mean age 57.5 years) with 164 malignant and 57 benign, taken from a radiological database that has been examined by a radiologist with more than 20 years of experience. Also, histopathological record data that had been examined by an oncologist with more than 20 years of experience. Mammograms were taken from January 2022 to June 2022. In all, 221 mammograms consisting of 164 malignant and 57 benign were used as SVM method training, and 20 mammograms consisting of 10 malignant and 10 benign were used to test the performance of the SVM method. It was then evaluated using pathology results as the gold standard. Results: Benign had a significantly lower deviation (an average of 29.2661230 ± 10.14916673) than malignant (an average of 33.1841234 ± 11.70238757). The SVM method performance value obtained the values of TP, FP, TN, FN, accuracy, sensitivity, Specificity, and Precision, respectively 7,7, 3, 3, 50%, 70%, 30%, and 50%. Conclusion: A proper performance to distinguish benign and malignant can be obtained using the physical deviation parameters with the SVM classification approach. However, these findings should be proven in larger datasets with different mammographic scanners. Our meta-analysis shows that the physical parameters and SVM have high sensitivity but low specificity. Of the nine physical parameters in the mammogram, only the parameter deviation was significant to distinguish between benign and malignant. The SVM method proved to be able to differentiate between benign and malignant.
Recently a new family of loss functions called smart error sums has been suggested. These loss functions account for correlations within experimental data and force modeled data to obey these correlations. As a result, multiplicative systematic errors of experimental data can be revealed and corrected. The smart error sums are based on 2D correlation analysis which is a comparably recent methodology for analyzing spectroscopic data that has found broad application. In this contribution we mathematically generalize and break down this methodology and the smart error sums to uncover the mathematic roots and simplify it to craft a general tool beyond spectroscopic modelling. This reduction also allows a simplified discussion about limits and prospects of this new method including one of its potential future uses as a sophisticated loss function in deep learning. To support its deployment, the work includes computer code to allow reproduction of the basic results.
Purpose
Proof‐of‐concept study of mapping renal blood flow vector field according to the inverse solution to a mass transport model of time resolved tracer‐labeled MRI data.
Theory and Methods
To determine tissue perfusion according to the underlying physics of spatiotemporal tracer concentration variation, the mass transport equation is integrated over a voxel with an approximate microvascular network for fitting time‐resolved tracer imaging data. The inverse solution to the voxelized transport equation provides the blood flow vector field, which is referred to as quantitative transport mapping (QTM). A numerical microvascular network modeling the kidney with computational fluid dynamics reference was used to verify the accuracy of QTM and the current Kety’s method that uses a global arterial input function. Multiple post‐label delay arterial spin labeling (ASL) of the kidney on seven subjects was used to assess QTM in vivo feasibility.
Results
Against the ground truth in the numerical model, the error in flow estimated by QTM (18.6%) was smaller than that in Kety’s method (45.7%, 2.5‐fold reduction). The in vivo kidney perfusion quantification by QTM (cortex: 443 ± 58 mL/100 g/min and medulla: 190 ± 90 mL/100 g/min) was in the range of that by Kety’s method (482 ± 51 mL/100 g/min in the cortex and 242 ± 73 mL/100 g/min in the medulla), and QTM provided better flow homogeneity in the cortex region.
Conclusions
QTM flow velocity mapping is feasible from multi‐delay ASL MRI data based on inverting the transport equation. In a numerical simulation, QTM with deconvolution in space and time provided more accurate perfusion quantification than Kety’s method with deconvolution in time only.
Arterial input function (AIF) is estimated from perfusion images as a basic curve for the following deconvolution process to calculate hemodynamic variables to evaluate vascular status of tissues. However, estimation of AIF is currently based on manual annotations with prior knowledge. We propose an automatic estimation of AIF in perfusion images based on a multi-stream 3D CNN, which combined spatial and temporal features together to estimate the AIF ROI. The model is trained by manual annotations. The proposed method was trained and tested with 100 cases of perfusion-weighted imaging. The result was evaluated by dice similarity coefficient, which reached 0.79. The trained model had a better performance than the traditional method. After segmentation of the AIF ROI, the AIF was calculated by the average of all voxels in the ROI. We compared the AIF result with the manual and traditional methods, and the parameters of further processing of AIF, such as time to the maximum of the tissue residue function (Tmax), relative cerebral blood flow, and mismatch volume, which are calculated in the Section Results. The result had a better performance, the average mismatch volume reached 93.32% of the manual method, while the other methods reached 85.04 and 83.04%. We have applied the method on the cloud platform, Estroke, and the local version of its software, NeuBrainCare, which can evaluate the volume of the ischemic penumbra, the volume of the infarct core, and the ratio of mismatch between perfusion and diffusion images to help make treatment decisions, when the mismatch ratio is abnormal.
The Prostate Imaging Reporting and Data System version 2 (PI-RADS v2) was developed with a consensus-based process using a combination of published data, and expert observations and opinions. In the short time since its release, numerous studies have validated the value of PI-RADS v2 but, as expected, have also identified a number of ambiguities and limitations, some of which have been documented in the literature with potential solutions offered. To address these issues, the PI-RADS Steering Committee, again using a consensus-based process, has recommended several modifications to PI-RADS v2, maintaining the framework of assigning scores to individual sequences and using these scores to derive an overall assessment category. This updated version, described in this article, is termed PI-RADS v2.1. It is anticipated that the adoption of these PI-RADS v2.1 modifications will improve inter-reader variability and simplify PI-RADS assessment of prostate magnetic resonance imaging even further. Research on the value and limitations on all components of PI-RADS v2.1 is strongly encouraged.
Purpose
We propose a novel methodology to integrate morphological and functional information of tumor‐associated vessels to assist in the diagnosis of suspicious breast lesions.
Theory and Methods
Ultrafast, fast, and high spatial resolution DCE‐MRI data were acquired on 15 patients with suspicious breast lesions. Segmentation of the vasculature from the surrounding tissue was performed by applying a Hessian filter to the enhanced image to generate a map of the probability for each voxel to belong to a vessel. Summary measures were generated for vascular morphology, as well as the inputs and outputs of vessels physically connected to the tumor. The ultrafast DCE‐MRI data was analyzed by a modified Tofts model to estimate the bolus arrival time, Ktrans (volume transfer coefficient), and vp (plasma volume fraction). The measures were compared between malignant and benign lesions via the Wilcoxon test, and then incorporated into a logistic ridge regression model to assess their combined diagnostic ability.
Results
A total of 24 lesions were included in the study (13 malignant and 11 benign). The vessel count, Ktrans, and vp showed significant difference between malignant and benign lesions (P = 0.009, 0.034, and 0.010, area under curve [AUC] = 0.76, 0.63, and 0.70, respectively). The best multivariate logistic regression model for differentiation included the vessel count and bolus arrival time (AUC = 0.91).
Conclusion
This study provides preliminary evidence that combining quantitative characterization of morphological and functional features of breast vasculature may provide an accurate means to diagnose breast cancer.
A new methodology based on tensor algebra that uses a higher order singular value decomposition to perform three-dimensional voxel reconstruction from a series of temporal images obtained using dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is proposed. Principal component analysis (PCA) is used to robustly extract the spatial and temporal image features and simultaneously de-noise the datasets. Tumour segmentation on enhanced scaled (ES) images performed using a fuzzy C-means (FCM) cluster algorithm is compared with that achieved using the proposed tensorial framework. The proposed algorithm explores the correlations between spatial and temporal features in the tumours. The multi-channel reconstruction enables improved breast tumour identification through enhanced de-noising and improved intensity consistency. The reconstructed tumours have clear and continuous boundaries; furthermore the reconstruction shows better voxel clustering in tumour regions of interest. A more homogenous intensity distribution is also observed, enabling improved image contrast between tumours and background, especially in places where fatty tissue is imaged. The fidelity of reconstruction is further evaluated on the basis of five new qualitative metrics. Results confirm the superiority of the tensorial approach. The proposed reconstruction metrics should also find future applications in the assessment of other reconstruction algorithms.
Delays between contrast agent (CA) arrival at the site of vascular input function (VIF) sampling and the tissue of interest affect dynamic contrast enhanced (DCE) MRI pharmacokinetic modelling. We investigate effects of altering VIF CA bolus arrival delays on liver DCE MRI perfusion parameters, propose an alternative approach to estimating delays and evaluate reproducibility.
Thirteen healthy volunteers (28.7 ± 1.9 years, seven males) underwent liver DCE MRI using dual-input single compartment modelling, with reproducibility (n = 9) measured at 7 days. Effects of VIF CA bolus arrival delays were assessed for arterial and portal venous input functions. Delays were pre-estimated using linear regression, with restricted free modelling around the pre-estimated delay. Perfusion parameters and 7 days reproducibility were compared using this method, freely modelled delays and no delays using one-way ANOVA. Reproducibility was assessed using Bland–Altman analysis of agreement.
Maximum percent change relative to parameters obtained using zero delays, were −31% for portal venous (PV) perfusion, +43% for total liver blood flow (TLBF), +3247% for hepatic arterial (HA) fraction, +150% for mean transit time and −10% for distribution volume. Differences were demonstrated between the 3 methods for PV perfusion (p = 0.0085) and HA fraction (p < 0.0001), but not other parameters. Improved mean differences and Bland–Altman 95% Limits-of-Agreement for reproducibility of PV perfusion (9.3 ml/min/100 g, ±506.1 ml/min/100 g) and TLBF (43.8 ml/min/100 g, ±586.7 ml/min/100 g) were demonstrated using pre-estimated delays with constrained free modelling.
CA bolus arrival delays cause profound differences in liver DCE MRI quantification. Pre-estimation of delays with constrained free modelling improved 7 days reproducibility of perfusion parameters in volunteers.
Purpose:
Fitting tracer kinetic models using linear methods is much faster than using their nonlinear counterparts, although this comes often at the expense of reduced accuracy and precision. The aim of this study was to derive and compare the performance of the linear compartmental tissue uptake (CTU) model with its nonlinear version with respect to their percentage error and precision.
Theory and methods:
The linear and nonlinear CTU models were initially compared using simulations with varying noise and temporal sampling. Subsequently, the clinical applicability of the linear model was demonstrated on 14 patients with locally advanced cervical cancer examined with dynamic contrast-enhanced magnetic resonance imaging.
Results:
Simulations revealed equal percentage error and precision when noise was within clinical achievable ranges (contrast-to-noise ratio >10). The linear method was significantly faster than the nonlinear method, with a minimum speedup of around 230 across all tested sampling rates. Clinical analysis revealed that parameters estimated using the linear and nonlinear CTU model were highly correlated (ρ ≥ 0.95).
Conclusion:
The linear CTU model is computationally more efficient and more stable against temporal downsampling, whereas the nonlinear method is more robust to variations in noise. The two methods may be used interchangeably within clinical achievable ranges of temporal sampling and noise. Magn Reson Med, 2016. © 2016 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc. on behalf of International Society for Magnetic Resonance in Medicine. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
To assess the reproducibility of brain quantitative susceptibility mapping (QSM) in healthy subjects and in patients with multiple sclerosis (MS) on 1.5 and 3T scanners from two vendors.
Ten healthy volunteers and 10 patients were scanned twice on a 3T scanner from one vendor. The healthy volunteers were also scanned on a 1.5T scanner from the same vendor and on a 3T scanner from a second vendor. Similar imaging parameters were used for all scans. QSM images were reconstructed using a recently developed nonlinear morphology-enabled dipole inversion (MEDI) algorithm with L1 regularization. Region-of-interest (ROI) measurements were obtained for 20 major brain structures. Reproducibility was evaluated with voxel-wise and ROI-based Bland-Altman plots and linear correlation analysis.
ROI-based QSM measurements showed excellent correlation between all repeated scans (correlation coefficient R ≥ 0.97), with a mean difference of less than 1.24 ppb (healthy subjects) and 4.15 ppb (patients), and 95% limits of agreements of within -25.5 to 25.0 ppb (healthy subjects) and -35.8 to 27.6 ppb (patients). Voxel-based QSM measurements had a good correlation (0.64 ≤ R ≤ 0.88) and limits of agreements of -60 to 60 ppb or less.
Brain QSM measurements have good interscanner and same-scanner reproducibility for healthy and MS subjects, respectively, on the systems evaluated in this study. J. Magn. Reson. Imaging 2015.
© 2015 Wiley Periodicals, Inc.
Arterial input function (AIF) plays an important role in the quantification of cerebral hemodynamics. The purpose of this study was to select the best reproducible clustering method for AIF detection by comparing three algorithms reported previously in terms of detection accuracy and computational complexity.
First, three reproducible clustering methods, normalized cut (Ncut), hierarchy (HIER), and fast affine propagation (FastAP), were applied independently to simulated data which contained the true AIF. Next, a clinical verification was performed where 42 subjects participated in dynamic susceptibility contrast MRI (DSC-MRI) scanning. The manual AIF and AIFs based on the different algorithms were obtained. The performance of each algorithm was evaluated based on shape parameters of the estimated AIFs and the true or manual AIF. Moreover, the execution time of each algorithm was recorded to determine the algorithm that operated more rapidly in clinical practice.
In terms of the detection accuracy, Ncut and HIER method produced similar AIF detection results, which were closer to the expected AIF and more accurate than those obtained using FastAP method; in terms of the computational efficiency, the Ncut method required the shortest execution time.
Ncut clustering appears promising because it facilitates the automatic and robust determination of AIF with high accuracy and efficiency.
Figure optionsDownload full-size imageDownload high-quality image (65 K)Download as PowerPoint slide
Induction of tumor angiogenesis is among the hallmarks of cancer and a driver of metastatic cascade initiation. Recent advances in high-resolution imaging enable highly detailed three-dimensional geometrical representation of the whole-tumor microvascular architecture. This enormous increase in complexity of image-based data necessitates the application of informatics methods for the analysis, mining and reconstruction of these spatial graph data structures. We present a novel methodology that combines ex-vivo high-resolution micro-computed tomography imaging data with a bioimage informatics algorithm to track and reconstruct the whole-tumor vasculature of a human breast cancer model. The reconstructed tumor vascular network is used as an input of a computational model that estimates blood flow in each segment of the tumor microvascular network. This formulation involves a well-established biophysical model and an optimization algorithm that ensures mass balance and detailed monitoring of all the vessels that feed and drain blood from the tumor microvascular network. Perfusion maps for the whole-tumor microvascular network are computed. Morphological and hemodynamic indices from different regions are compared to infer their role in overall tumor perfusion.
Phantoms for dynamic contrast enhanced (DCE) imaging modalities such as DCE computed tomography (DCE-CT) and DCE magnetic resonance imaging (DCE-MRI) are valuable tools for evaluating and comparing imaging systems. It is important for the contrast-agent distribution within the phantom to possess a time dependence that replicates a curve observed clinically, known as the 'tumor-enhancement curve'. It is also important for the concentration field within the lesion to be as uniform as possible. This study demonstrates how computational fluid dynamics (CFD) can be applied to achieve these goals within design constraints. The distribution of the contrast agent within the simulated phantoms was investigated in relation to the influence of three factors of the phantom design. First, the interaction between the inlets and the uniformity of the contrast agent within the phantom was modeled. Second, pumps were programmed using a variety of schemes and the resultant dynamic uptake curves were compared to tumor-enhancement curves obtained from clinical data. Third, the effectiveness of pulsing the inlet flow rate to produce faster equilibration of the contrast-agent distribution was quantified. The models employed a spherical lesion and design constraints (lesion diameter, inlet-tube size and orientation, contrast-agent flow rates and fluid properties) taken from a recently published DCE-MRI phantom study. For DCE-MRI in breast cancer detection, where the target tumor-enhancement curve varies on the scale of hundreds of seconds, optimizing the number of inlet tubes and their orientation was found to be adequate for attaining concentration uniformity and reproducing the target tumor-enhancement curve. For DCE-CT in liver tumor detection, where the tumor-enhancement curve varies on a scale of tens of seconds, the use of an iterated inlet condition (programmed into the pump) enabled the phantom to reproduce the target tumor-enhancement curve within a few per cent beyond about 6 s of wash-in. This time was cut in half by the final CFD-derived strategy of flow pulsing. Driving the pump with a 25% duty cycle pulsatile waveform produced a nearly uniform concentration in the phantom in just a few seconds under typical conditions. Comparisons with published x-ray measurements using tumor-enhancement curves for both benign and malignant breast lesions showed a difference of approximately 4% between the CFD predictions and measurements of the contrast-agent concentration averaged over the lesion volume. The techniques derived using CFD optimization can be used in future phantom designs, including as starting points for future CFD phantom studies employing new lesion geometries and tumor-enhancement curves.
The semi-quantitative analysis of the time-intensity curves in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a limited specificity due to overlapping enhancement patterns after gadolinium administration. With the advances in technology and faster sequences, imaging of the entire breast can be done in a few seconds, which allows measuring the transit of contrast (transfer constant: K(trans)) through the vascular bed at capillary level that reflects quantitative measure of porosity/permeability of tumor vessels.
Our study aims to evaluate the pharmacokinetic parameter K(trans) for enhancing breast lesions and correlate it with histopathology, and assess accuracy, sensitivity, and specificity of this parameter in discriminating benign and malignant breast lesions.
One hundred and fifty-one women with 216 histologically proved enhancing breast lesions underwent high temporal resolution DCE-MRI for the early dynamic analysis for calculation of pharmacokinetic parameters (K(trans)) using standard two compartment model. The calculated values of K(trans) were correlated with histopathology to calculate the sensitivity, specificity, and accuracy.
Receiver operating characteristic (ROC) curve analysis revealed a mean K(trans) value of 0.56, which reliably distinguished benign and malignant breast lesions with a sensitivity of 91.1% and specificity of 90.3% with an overall accuracy of 89.3%. The area under curve (AUC) was 0.907.
K(trans) is a reliable quantitative parameter for characterizing benign and malignant lesions in routine DCE-MRI of breasts.
The connection between fluid flow and optical flow is explored in typical flow visualizations to provide a rational foundation for application of the optical flow method to image-based fluid velocity measurements. The projected-motion equations are derived, and the physics-based optical flow equation is given. In general, the optical flow is proportional to the path-averaged velocity of fluid or particles weighted with a relevant field quantity. The variational formulation and the corresponding Euler–Lagrange equation are given for optical flow computation. An error analysis for optical flow computation is provided, which is quantitatively examined by simulations on synthetic grid images. Direct comparisons between the optical flow method and the correlation-based method are made in simulations on synthetic particle images and experiments in a strongly excited turbulent jet.
A method to simultaneously estimate the arterial input function (AIF) and pharmacokinetic model parameters from dynamic contrast-enhanced (DCE)-MRI data was developed. This algorithm uses a parameterized functional form to model the AIF and k-means clustering to classify tissue time-concentration measurements into a set of characteristic curves. An iterative blind estimation algorithm alternately estimated parameters for the input function and the pharmacokinetic model. Computer simulations were used to investigate the algorithm's sensitivity to noise and initial estimates. In 12 patients with sarcomas, pharmacokinetic parameter estimates were compared with "truth" obtained from model regression using a measured AIF. When arterial voxels were included in the blind estimation algorithm, the resulting AIF was similar to the measured input function. The "true" K(trans) values in tumor regions were not significantly different than the estimated values, 0.99 +/- 0.41 and 0.86 +/- 0.40 min(-1), respectively, P = 0.27. "True" k(ep) values also matched closely, 0.70 +/- 0.24 and 0.65 +/- 0.25 min(-1), P = 0.08. When only tissue curves free of significant vascular contribution are used (v(p) < 0.05), the resulting AIF showed substantial delay and dispersion consistent with a more local AIF such as has been observed in dynamic susceptibility contrast imaging in the brain.
An automatic segmentation technique has been developed and applied to two renal micro-computer tomography (CT) images. With the use of a 20-microm voxel resolution image, the arterial and venous trees were segmented for the rat renal vasculature, distinguishing resolving vessels down to 30 microm in radius. A higher resolution 4-microm voxel image of a renal vascular subtree, with vessel radial values down to 10 microm, was segmented. Strahler ordering was applied to each subtree using an iterative scheme developed to integrate information from the two segmented models to reconstruct the complete topology of the entire vascular tree. An error analysis of the assigned orders quantified the robustness of the ordering process for the full model. Radial, length, and connectivity data of the complete arterial and venous trees are reported by order. Substantial parallelism is observed between individual arteries and veins, and the ratio of parallel vessel radii is quantified via a power law. A strong correlation with Murray's Law was established, providing convincing evidence of the "minimum work" hypothesis. Results were compared with theoretical branch angle formulations, based on the principles of "minimum shear force," were inconclusive. Three-dimensional reconstructions of renal vascular trees collected are made freely available for further investigation into renal physiology and modeling studies.
The overall goal of this study is to employ quantitative magnetic resonance imaging (MRI) data to constrain a patient-specific, computational fluid dynamics (CFD) model of blood flow and interstitial transport in breast cancer. We develop image processing methodologies to generate tumor-related vasculatureinterstitium geometry and realistic material properties, using dynamic contrast enhanced MRI (DCE-MRI) and diffusion weighted MRI (DW-MRI) data. These data are used to constrain CFD simulations for determining the tumorassociated blood supply and interstitial transport characteristics unique to each patient. We then perform a proof-of-principle statistical comparison between these hemodynamic characteristics in 11 malignant and 5 benign lesions from 12 patients. Significant differences between groups (i.e., malignant versus benign) were observed for the median of tumor-associated interstitial flow velocity (P = 0.028), and the ranges of tumor-associated blood pressure (P = 0.016) and vascular extraction rate (P = 0.040). The implication is that malignant lesions tend to have larger magnitude of interstitial flow velocity, and higher heterogeneity in blood pressure and vascular extraction rate. Multivariable logistic models based on combinations of these hemodynamic data achieved excellent differentiation between malignant and benign lesions with an area under the receiver operator characteristic curve of 1.0, sensitivity of 1.0, and specificity of 1.0. This imagebased model system is a fundamentally new way to map flow and pressure fields related to breast tumors using only non-invasive, clinically available imaging data and established laws of fluid mechanics. Furthermore, the results provide preliminary evidence for this methodology’s utility for the quantitative characterization of breast cancer.
Comparative preliminary analysis of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) data collected in the International Breast MR Consortium 6883 multicenter trial was performed to distinguish benign and malignant breast tumors. Prebiopsy DCE-MRI data from 45 patients with suspicious breast lesions were obtained. Semiquantitative mean signal-enhancement ratio ([Formula: see text]) was calculated for all lesions, and quantitative pharmacokinetic, parameters [Formula: see text], [Formula: see text], and [Formula: see text], were calculated for the subset with available [Formula: see text] maps ([Formula: see text]). Diagnostic performance was estimated for DCE-MRI parameters and compared to standard clinical MRI assessment. Quantitative and semiquantitative metrics discriminated benign and malignant lesions, with receiver operating characteristic area under the curve (AUC) values of 0.71, 0.70, and 0.82 for [Formula: see text], [Formula: see text], and [Formula: see text], respectively ([Formula: see text]). At equal 94% sensitivity, the specificity and positive predictive value of [Formula: see text] (53% and 63%, respectively) and Ktrans (42% and 58%) were higher than clinical MRI assessment (32% and 54%). A multivariable model combining [Formula: see text] and clinical MRI assessment had an AUC value of 0.87. Quantitative pharmacokinetic and semiquantitative analyses of DCE-MRI improves discrimination of benign and malignant breast tumors, with our findings suggesting higher diagnostic accuracy using [Formula: see text]. [Formula: see text] has potential to help reduce unnecessary biopsies resulting from routine breast imaging.
Purpose:
To propose an efficient algorithm to perform dual input compartment modeling for generating perfusion maps in the liver.
Methods:
We implemented whole field-of-view linear least squares (LLS) to fit a delay-compensated dual-input single-compartment model to very high temporal resolution (four frames per second) contrast-enhanced 3D liver data, to calculate kinetic parameter maps. Using simulated data and experimental data in healthy subjects and patients, whole-field LLS was compared with the conventional voxel-wise nonlinear least-squares (NLLS) approach in terms of accuracy, performance, and computation time.
Results:
Simulations showed good agreement between LLS and NLLS for a range of kinetic parameters. The whole-field LLS method allowed generating liver perfusion maps approximately 160-fold faster than voxel-wise NLLS, while obtaining similar perfusion parameters.
Conclusions:
Delay-compensated dual-input liver perfusion analysis using whole-field LLS allows generating perfusion maps with a considerable speedup compared with conventional voxel-wise NLLS fitting. Magn Reson Med, 2017. © 2017 International Society for Magnetic Resonance in Medicine.
Purpose:
To develop a portable perfusion phantom and validate its utility in quantitative dynamic contrast-enhanced magnetic resonance imaging of the abdomen.
Methods:
A portable perfusion phantom yielding a reproducible contrast enhancement curve (CEC) was developed. A phantom package including perfusion and static phantoms were imaged simultaneously with each of three healthy human volunteers in two different 3T MR scanners. Look-up tables correlating reference (known) contrast concentrations with measured ones were created using either the static or perfusion phantom. Contrast maps of image slices showing four organs (liver, spleen, pancreas, and paravertebral muscle) were generated before and after data correction using the look-up tables. The contrast concentrations at 4.5 minutes after dosing in each of the four organs were averaged for each volunteer. The mean contrast concentrations (4 organs x 3 volunteers = 12) were compared for the two scanners, and the intra-class correlation coefficient (ICC) was calculated. Also, the ICC of the mean K(trans) values between the two scanners was calculated before and after data correction.
Results:
The repeatability coefficient of CECs of perfusion phantom was higher than 0.997 in all measurements. The ICC of the tissue contrast concentrations between the two scanners was 0.693 before correction, but increased to 0.974 after correction using the look-up tables (LUTs) of perfusion phantom. However, the ICC was not increased after correction using static phantom (ICC: 0.617). Similarly, the ICC of the K(trans) values was 0.899 before correction, but increased to 0.996 after correction using perfusion phantom LUTs. The ICC of the K(trans) values, however, was not increased when static phantom LUTs were used (ICC: 0.866).
Conclusions:
The perfusion phantom reduced variability in quantifying contrast concentration and K(trans) values of human abdominal tissues across different MR units, but static phantom did not. The perfusion phantom has the potential to facilitate multi-institutional clinical trials employing quantitative DCE-MRI to evaluate various abdominal malignancies. This article is protected by copyright. All rights reserved.
Purpose:
Kinetic parameters derived from dynamic contrast-enhanced MRI (DCE-MRI) were suggested as a possible instrument for multi-parametric lesion characterization, but have not found their way into clinical practice yet due to inconsistent results. The quantification is heavily influenced by the definition of an appropriate arterial input functions (AIF). Regarding brain tumor DCE-MRI, there are currently several co-existing methods to determine the AIF frequently including different brain vessels as sources. This study quantitatively and qualitatively analyzes the impact of AIF source selection on kinetic parameters derived from commonly selected AIF source vessels compared to a population-based AIF model.
Material and methods:
74 patients with brain lesions underwent 3D DCE-MRI. Kinetic parameters [transfer constants of contrast agent efflux and reflux K(trans) and kep, their ratio, ve, to measure extravascular-extracellular volume fraction and plasma volume fraction vp] were determined using extended Tofts model in 821 ROI from 4 AIF sources [the internal carotid artery (ICA), the closest artery to the lesion, the superior sagittal sinus (SSS), the population-based Parker model]. The effect of AIF source alteration on kinetic parameters was evaluated by tissue type selective intra-class correlation (ICC) and capacity to differentiate gliomas by WHO grade [area under the curve analysis (AUC)].
Results:
Arterial AIF more often led to implausible ve >100% values (p<0.0001). AIF source alteration rendered different absolute kinetic parameters (p<0.0001), except for kep. ICC between kinetic parameters of different AIF sources and tissues were variable (0.08-0.87) and only consistent >0.5 between arterial AIF derived kinetic parameters. Differentiation between WHO III and II glioma was exclusively possible with vp derived from an AIF in the SSS (p=0.03; AUC 0.74).
Conclusion:
The AIF source has a significant impact on absolute kinetic parameters in DCE-MRI, which limits the comparability of kinetic parameters derived from different AIF sources. The effect is also tissue-dependent. The SSS appears to be the best choice for AIF source vessel selection in brain tumor DCE-MRI as it exclusively allowed for WHO grades II/III and III/IV glioma distinction (by vp) and showed the least number of implausible ve values.
Purpose:
To assess the diagnostic utility of contrast kinetic analysis for breast lesions and background parenchyma of women undergoing MRI-guided biopsies, for whom standard clinical analysis had failed to separate benign and malignant lesions.
Materials and methods:
This study included 115 women who had indeterminate lesions based on routine diagnostic breast MRI exams and underwent an MRI (3 Tesla) -guided biopsy of one or more lesions suspicious for breast cancer. Breast dynamic contrast-enhanced (DCE)-MRI was performed using a radial stack-of-stars three-dimensional spoiled gradient echo pulse sequence and modified k-space weighted image contrast image reconstruction. Contrast kinetic model analysis was conducted to characterize the contrast enhancement patterns measured in lesions and background parenchyma (BP). The transfer rate (K(trans) ), interstitial volume fraction (ve ), and vascular volume fraction (vp ) estimated from the lesion and BP were used to separate malignant from benign lesions.
Results:
The patients with malignant lesions had significantly (P < 0.05) higher median lesion-K(trans) (0.081 min(-1) ), higher median BP-K(trans) (0.032 min(-1) ), and BP-vp (0.020) than those without malignant lesions (0.056 min(-1) , 0.017 min(-1) and 0.012, respectively). The area under the receiver operating characteristic curve (AUC) of the BP-K(trans) (0.687) was highest among the single parameters and higher than that of the lesion-K(trans) (0.664). The combined logistic regression model of lesion-K(trans) , lesion-ve , BP-K(trans) , BP-ve , and BP-vp had the highest AUC of 0.730.
Conclusion:
Our results suggest that the contrast kinetic analysis of DCE-MRI data can be used to differentiate the malignant lesions from the benign and high-risk lesions among the indeterminate breast lesions recommended for MRI-guided biopsy exams.
Level of evidence:
3 J. MAGN. RESON. IMAGING 2017;45:1385-1393.
Objectives:
To evaluate the performance of six models of population Arterial Input Function (AIF) in the setting of primary breast cancer and neoadjuvant chemotherapy (NAC). The ability to fit patient DCE-MRI data, provide physiological plausible data and detect pathological response are assessed.
Materials and methods:
Quantitative DCE-MRI parameters were calculated for 27 patients at baseline and after 2 cycles of NAC for 6 AIFs. Pathologic complete response (pCR) detection was compared to change in these parameters from a reproduction cohort of 12 patients using a Bland-Altman approach and receiver operator characteristic (ROC) analysis.
Results:
There were fewer fit failures pre NAC for all models with the Modified Fritz-Hansen having the fewest pre NAC (3.6%) and post NAC (18.8%), contrasting with the Femoral Artery AIF (19.4% and 43.3% respectively). Median K(trans) values were greatest for the Weinmann function and also showed greatest reductions with treatment (-68%). Reproducibility (r) was the lowest for the Weinmann function (r = -49.7%) with the other AIFs ranging from r -27.8 to r -39.2%.
Conclusion:
Using the best performing AIF is essential to maximise the utility of quantitative DCE-MRI parameters in predicting response to NAC treatment. Applying our criteria the Modified Fritz-Hansen and Cosine Bolus approximated Parker AIF models performed best. The Fritz-Hansen and biexponential approximated Parker AIFs performed less well and the Weinmann and femoral artery AIFs are not recommended. Advances in Knowledge: We demonstrate that using the most appropriate Arterial Input Function (AIF) can aid successful prediction of response to neoadjuvant chemotherapy in breast cancer.
The time-signal intensity curve (TIC) from dynamic contrast-enhanced magnetic resonance imaging (DCEMRI) reflects the hemodynamic features of a specific lesion. The TIC is obtained by repeated MRI scans after the injection of contrast agent; a qualified TIC usually takes 12 minutes to complete the scans. Temporal resolution is the main determinant; a higher temporal resolution means a smoother TIC.
PurposeThe purpose of this study is to develop a dynamic quantitative susceptibility mapping (QSM) technique with sufficient temporal resolution to map contrast agent concentration in cerebral perfusion imaging.Methods
The dynamic QSM used a multiecho three-dimensional (3D) spoiled gradient echo golden angle interleaved spiral sequence during contrast bolus injection. Four-dimensional (4D) space-time resolved magnetic field reconstruction was performed using the temporal resolution acceleration with constrained evolution reconstruction method. Deconvolution of the gadolinium-induced field was performed at each time point with the morphology enabled dipole inversion method to generate a 4D gadolinium concentration map, from which three-dimensional spatial distributions of cerebral blood volume and cerebral blood flow were computed.ResultsInitial in vivo brain imaging demonstrated the feasibility of using dynamic QSM for generating quantitative 4D contrast agent maps and imaging three-dimensional perfusion. The cerebral blood flow obtained with dynamic QSM agreed with that obtained using arterial spin labeling.Conclusion
Dynamic QSM can be used to perform 4D mapping of contrast agent concentration in contrast-enhanced magnetic resonance imaging. The perfusion parameters derived from this 4D contrast agent concentration map were in good agreement with those obtained using arterial spin labeling. Magn Reson Med, 2014. © 2014 Wiley Periodicals, Inc.
Dynamic imaging data are currently analyzed with a tracer-kinetic theory developed for individual time curves measured over whole organs. The assumption is that voxels represent isolated systems which all receive indicator through the same arterial inlet. This leads to well-known systematic errors, but also fails to exploit the spatial structure of the data. In this study, a more general theoretical framework is developed which makes full use of the specific structure of image data. The theory encodes the fact that voxels receive indicator from their immediate neighbors rather than from an upstream arterial input. This results in a tracer-kinetic field theory where the tissue parameters are functions of space which can be measured by analyzing the temporal and spatial patterns in the concentrations. The implications are evaluated through a number of field models for common tissue types. The key benefits of a tracer-kinetic field theory are that: 1) long-standing systematic errors can be corrected, specifically the issue of bolus dispersion and the contamination of large-vessel blood flow on tissue perfusion measurements; 2) additional tissue parameters can be measured that characterize convective or diffusive exchange between voxels; 3) the need to measure a separate arterial input function can be eliminated.
An algorithm is developed for the reconstruction of dynamic, gadolinium (Gd) bolus MR perfusion images of the human brain, based on quantitative susceptibility mapping (QSM).
The method is evaluated in five perfusion scans obtained from four different patients scanned at 3 Tesla, and compared with the conventional analysis based on changes in the transverse relaxation rate ΔR2 * and to theoretical predictions. QSM images were referenced to ventricular cerebrospinal fluid (CSF) for each dynamic of the perfusion sequence.
Images of cerebral blood flow and blood volume were successfully reconstructed from the QSM-analysis, and were comparable to those reconstructed using ΔR2 *. The magnitudes of the Gd-associated susceptibility effects in gray and white matter were consistent with theoretical predictions.
QSM-based analysis may have some theoretical advantages compared with ΔR2 *, including a simpler relationship between signal change and Gd concentration. However, disadvantages are its much lower contrast-to-noise ratio, artifacts due to respiration and other effects, and more complicated reconstruction methods. More work is required to optimize data acquisition protocols for QSM-based perfusion imaging. Magn Reson Med, 2014. © 2014 Wiley Periodicals, Inc.
Cerebral perfusion, also referred to as cerebral blood flow (CBF), is one of the most important parameters related to brain physiology and function. The technique of dynamic-susceptibility contrast (DSC) MRI is currently the most commonly used MRI method to measure perfusion. It relies on the intravenous injection of a contrast agent and the rapid measurement of the transient signal changes during the passage of the bolus through the brain. Central to quantification of CBF using this technique is the so-called arterial input function (AIF), which describes the contrast agent input to the tissue of interest. Due to its fundamental role, there has been a lot of progress in recent years regarding how and where to measure the AIF, how it influences DSC-MRI quantification, what artefacts one should avoid, and the design of automatic methods to measure the AIF. The AIF is also directly linked to most of the major sources of artefacts in CBF quantification, including partial volume effect, bolus delay and dispersion, peak truncation effects, contrast agent non-linearity, etc. While there have been a number of good review articles on DSC-MRI over the years, these are often comprehensive but, by necessity, with limited in-depth discussion of the various topics covered. This review article covers in greater depth the issues associated with the AIF and their implications for perfusion quantification using DSC-MRI.
The purpose of this study was to analyze background parenchymal enhancement (BPE) in the contralateral normal breast of cancer patients during the course of neoadjuvant chemotherapy (NAC). Forty-five subjects were analyzed. Each patient had three MRIs, one baseline (B/L) and two follow-up (F/U) studies. The fibroglandular tissue in the contralateral normal breast was segmented using a computer-assisted algorithm. Based on the segmented fibroglandular tissue, BPE was calculated. BPE measured in baseline (B/L) and follow-up (F/U) MR studies were compared. The baseline BPE was also correlated with age and compared between pre/peri-menopausal (<55years old) and post-menopausal women (≥55years old). The pre-treatment BPE measured in B/L MRI was significantly higher in women <55years old than in women ≥55years old (20.1%±7.4% vs. 12.1%±5.1%, p≤0.01). A trend of negative correlation between BPE and age was noted (r=-0.29). In women <55years old, BPE at F/U-1 (18.8%±6.9%) was decreased compared to B/L, and was further decreased in F/U-2 (13.3%±5.7%) which was significant compared to B/L and F/U-1. In women ≥55years old, no significant difference was noted in any paired comparison among B/L, F/U-1 and F/U-2 MRI. A higher baseline BPE was associated with a greater reduction of BPE in F/U-2 MRI (r=0.73). Our study showed that younger women tended to have higher BPE than older women. BPE was significantly decreased in F/U-2 MRI after NAC in women <55years old. The reduction in BPE was most likely due to the ovarian ablation induced by chemotherapeutic agents.
The growth and dissemination of tumors rely on an altered vascular network, which supports their survival and expansion and provides accessibility to the vasculature and a route of transport for metastasizing tumor cells. The remodeling of vascular structures through generation of new vessels (for example, via tumor angiogenesis) is a well studied, even if still quite poorly understood, process in human cancer. Antiangiogenic therapies have provided insight into the contribution of angiogenesis to the biology of human tumors, yet have also revealed the ease with which resistance to antiangiogenic drugs can develop, presumably involving alterations to vascular signaling mechanisms. Furthermore, cellular and/or molecular changes to pre-existing vessels could represent subtle pre-metastatic alterations to the vasculature, which are important for cancer progression. These changes, and associated molecular markers, may forecast the behavior of individual tumors and contribute to the early detection, diagnosis and prognosis of cancer. This review, which primarily focuses on the blood vasculature, explores current knowledge of how tumor vessels can be remodeled, and the cellular and molecular events responsible for this process.Oncogene advance online publication, 5 August 2013; doi:10.1038/onc.2013.304.
Purpose:
To test the reproducibility of model-derived quantitative and semiquantitative pharmacokinetic parameters among various commercially available perfusion analysis solutions for dynamic contrast material-enhanced (DCE) magnetic resonance (MR) imaging.
Materials and methods:
The study was institutional review board approved and HIPAA compliant, with waiver of informed consent granted. The study group consisted of 15 patients (mean age, 44 years; range, 28-60 years), with 15 consecutive 1.5-T DCE MR imaging studies performed between October 1, 2010, and December 27, 2010, prior to uterine fibroid embolization. Studies were conducted by using variable-flip-angle T1 mapping and four-dimensional, time-resolved MR angiography with interleaved stochastic trajectories. Images from all DCE MR imaging studies were postprocessed with four commercially available perfusion analysis solutions by using a Tofts and Kermode model paradigm. Five observers measured pharmacokinetic parameters (volume transfer constant [K(trans)], v(e) [extracellular extravascular volume fraction], k(ep)[K(trans)/v(e)], and initial area under the gadolinium curve [iAUGC]) three times for each imaging study with each perfusion analysis solution (between March 13, 2011, and September 8, 2011) by using two different region-of-interest methods, resulting in 1800 data points.
Results:
After normalization of data output, significant differences in mean values were found for the majority of perfusion analysis solution combinations. The within-subject coefficient of variation among perfusion analysis solutions was 48.3%-68.8% for K(trans), 37.2%-60.3% for k(ep), 27.7%-74.1% for v(e), and 25.1%-61.2% for iAUGC. The intraclass correlation coefficient revealed only poor to moderate consistency among pairwise perfusion analysis solution comparisons (K(trans), 0.33-0.65; k(ep), 0.02-0.81; v(e), -0.03 to 0.72; and iAUGC, 0.47-0.78).
Conclusion:
A considerable variability for DCE MR imaging pharmacokinetic parameters (K(trans), k(ep), v(e), iAUGC) was found among commercially available perfusion analysis solutions. Therefore, clinical comparability across perfusion analysis solutions is currently not warranted. Agreement on a postprocessing standard is paramount prior to establishing DCE MR imaging as a widely incorporated biomarker.
An ECG-triggered magnetization-prepared segmented 3D fast gradient echo sequence was developed to perform pulmonary arterial MR angiography. A selective inversion recovery pulse was used in the magnetization preparation to suppress venous vasculature. A real-time gating technique based on navigator echoes was implemented to reduce respiration effects. Pencil-beam navigator echoes were acquired immediately before and after the readout train and processed in real-time to dynamically measure the diaphragm position, which was used to control data acquisition with an accept-or-reject-reacquire logic. In a study of 10 volunteers, a gated 3D acquisition with 28 slices required on average approximately 4 min of acquisition time, and six to seven segmental arteries related to the interlobar trunk of the pulmonary artery were depicted. The use of SIR pulse reduced venous signal by 99%. The gated acquisitions were superior to the ungated acquisitions (n = 10, P < 0.005). The real-time navigator gating technique is effective for reduction of respiration effects and thereby makes high resolution 3D MRA of the pulmonary arteries feasible.
Retrospective adaptive motion correction (AMC) was developed for reducing effects of residual respiration in real-time navigator-gated three-dimensional (3D) coronary magnetic resonance (MR) angiography. In both motion phantom and in vivo experiments, AMC improved image sharpness of coronary arteries. This navigator-based technique combining adaptive correction and real-time gating is potentially an efficient and effective motion reduction method for 3D coronary MR angiography. J. Magn. Reson. Imaging 2000;11:208–214. © 2000 Wiley-Liss, Inc.
Time-resolved imaging is crucial for the accurate diagnosis of liver lesions. Current contrast enhanced liver magnetic resonance imaging acquires a few phases in sequential breath-holds. The image quality is susceptible to bolus timing errors, which could result in missing the critical arterial phase. This impairs the detection of malignant tumors that are supplied primarily by the hepatic artery. In addition, the temporal resolution may be too low to reliably separate the arterial phase from the portal venous phase. In this study, a method called temporal resolution acceleration with constrained evolution reconstruction was developed with three-dimensional volume coverage and high-temporal frame rate. Data is acquired using a stack of spirals sampling trajectory combined with a golden ratio view order using an eight-channel coil array. Temporal frames are reconstructed from vastly undersampled data sets using a nonlinear inverse algorithm assuming that the temporal changes are small at short time intervals. Numerical and phantom experimental validation is presented. Preliminary in vivo results demonstrated high spatial resolution dynamic three-dimensional images of the whole liver with high frame rates, from which numerous subarterial phases could be easily identified retrospectively. Magn Reson Med, 2012. © 2012 Wiley Periodicals, Inc.
Compartmental tracer kinetic models currently used for analysis of dynamic contrast-enhanced MRI data yield poor fittings or parameter values that are unphysiological in necrotic regions of the tumor, as these models only describe microcirculation in perfused tissue. In this study, we explore the use of Fick's law of diffusion as an alternative method for analysis of dynamic contrast-enhanced MRI data in the necrotic regions. Xenografts of various human cancer cell lines were implanted in 14 mice that were subjected to dynamic contrast-enhanced MRI performed using a spoiled gradient recalled sequence. Tracer concentration was estimated using the variable flip angle technique. Poorly perfused and necrotic tumor regions exhibiting delayed and slow enhancement were identified using a k-means clustering algorithm. Tracer behavior in necrotic regions was shown to be consistent with Fick's diffusion equation and the in vivo gadolinium diffusivity was estimated to be 2.08 (±0.88) × 10(-4) mm(2) /s. This study proposes the use of gadolinium diffusivity as an alternative parameter for quantifying tracer transport within necrotic tumor regions. Magn Reson Med, 2012. © 2012 Wiley Periodicals, Inc.
To examine the variability in the qualitative and quantitative results of computed tomographic (CT) perfusion imaging generated from identical source data of stroke patients by using commercially available software programs provided by various CT manufacturers.
Institutional review board approval and informed consent were obtained. CT perfusion imaging data of 10 stroke patients were postprocessed by using five commercial software packages, each of which had a different algorithm: singular-value decomposition (SVD), maximum slope (MS), inverse filter (IF), box modulation transfer function (bMTF), and by using custom-made original software with standard (sSVD) and block-circulant (bSVD) SVD methods. Areas showing abnormalities in cerebral blood flow (CBF), mean transit time (MTT), and cerebral blood volume (CBV) were compared with each other and with the final infarct areas. Differences among the ratios of quantitative values in the final infarct areas and those in the unaffected side were also examined.
The areas with CBF or MTT abnormalities and the ratios of these values significantly varied among software, while those of CBV were stable. The areas with CBF or MTT abnormalities analyzed by using SVD or bMTF corresponded to those obtained with delay-sensitive sSVD, but overestimated the final infarct area. The values obtained from software by using MS or IF corresponded well with those obtained from the delay-insensitive bSVD and the final infarct area. Given the similarities between CBF and MTT, all software were separated in two groups (ie, sSVD and bSVD). The ratios of CBF or MTTs correlated well within both groups, but not across them.
CT perfusion imaging maps were significantly different among commercial software even when using identical source data, presumably because of differences in tracer-delay sensitivity.
The diagnosis of many neurologic diseases benefits from the ability to quantitatively assess iron in the brain. Paramagnetic iron modifies the magnetic susceptibility causing magnetic field inhomogeneity in MRI. The local field can be mapped using the MR signal phase, which is discarded in a typical image reconstruction. The calculation of the susceptibility from the measured magnetic field is an ill-posed inverse problem. In this work, a bayesian regularization approach that adds spatial priors from the MR magnitude image is formulated for susceptibility imaging. Priors include background regions of known zero susceptibility and edge information from the magnitude image. Simulation and phantom validation experiments demonstrated accurate susceptibility maps free of artifacts. The ability to characterize iron content in brain hemorrhage was demonstrated on patients with cavernous hemangioma. Additionally, multiple structures within the brain can be clearly visualized and characterized. The technique introduces a new quantitative contrast in MRI that is directly linked to iron in the brain.
The type of contrast enhancement kinetic curve (i.e., persistently enhancing, plateau, or washout) seen on dynamic contrast-enhanced MRI (DCE-MRI) of the breast is predictive of malignancy. Qualitative estimates of the type of curve are most commonly used for interpretation of DCE-MRI. The purpose of this study was to compare qualitative and quantitative methods for determining the type of contrast enhancement kinetic curve on DCE-MRI.
Ninety-six patients underwent breast DCE-MRI. The type of DCE-MRI kinetic curve was assessed qualitatively by three radiologists on two occasions. For quantitative assessment, the slope of the washout curve was calculated. Kappa statistics were used to determine inter- and intraobserver agreement for the qualitative method. Matched sample tables, the McNemar test, and receiver operating characteristic (ROC) curve statistics were used to compare quantitative versus qualitative methods for establishing or excluding malignancy.
Seventy-eight lesions (77.2%) were malignant and 23 (22.8%) were benign. For the qualitative assessment, the intra- and interobserver agreement was good (kappa = 0.76-0.88), with an area under the ROC curve (AUC) of 0.73-0.77. For the quantitative method, the highest AUC was 0.87, reflecting significantly higher diagnostic accuracies compared with qualitative assessment (p < 0.01 for the difference between the two methods).
Quantitative assessment of the type of contrast enhancement kinetic curve on breast DCE-MRI resulted in significantly higher diagnostic performance for establishing or excluding malignancy compared with assessment based on the standard qualitative method.
For pharmacokinetic modeling of tissue physiology, there is great interest in measuring the arterial input function (AIF) from dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) using paramagnetic contrast agents. Due to relaxation effects, the measured signal is a nonlinear function of the injected contrast agent concentration and depends on sequence parameters, system calibration, and time-of-flight effects, making it difficult to accurately measure the AIF during the first pass. Paramagnetic contrast agents also affect susceptibility and modify the magnetic field in proportion to their concentration. This information is contained in the MR signal phase which is discarded in a typical image reconstruction. However, quantifying AIF through contrast agent susceptibility induced phase changes is made difficult by the fact that the induced magnetic field is nonlocal and depends upon the contrast agent spatial distribution and thus on organ and vessel shapes. In this article, the contrast agent susceptibility was quantified through inversion of magnetic field shifts using a piece-wise constant model. Its feasibility is demonstrated by a determination of the AIF from the susceptibility-induced field changes of an intravenous bolus. After in vitro validation, a time-resolved two-dimensional (2D) gradient echo scan, triggered to diastole, was performed in vivo on the aortic arch during a bolus injection of 0.1 mmol/kg Gd-DTPA. An approximate geometrical model of the aortic arch constructed from the magnitude images was used to calculate the spatial variation of the field associated with the bolus. In 14 subjects, Gd concentration curves were measured dynamically (one measurement per heart beat) and indirectly validated by independent 2D cine phase contrast flow rate measurements. Flow rate measurements using indicator conservation with this novel quantitative susceptibility imaging technique were found to be in good agreement with those obtained from the cine phase contrast measurements in all subjects. Contrary to techniques that rely on intensity, the accuracy of this signal phase based method is insensitive to factors influencing signal intensity such as flip angle, coil sensitivity, relaxation changes, and time-of-flight effects extending the range of pulse sequences and contrast doses for which quantitative DCE-MRI can be applied.
Dynamic Contrast Enhanced MRI (DCE-MRI) is today one of the most popular methods for tumor assessment. Several pharmacokinetic models have been proposed to analyze DCE-MRI. Most of them depend on an accurate arterial input function (AIF). We propose an automatic and versatile method to determine the AIF. The method has two stages, detection and segmentation, incorporating knowledge about artery structure, fluid kinetics, and the dynamic temporal property of DCE-MRI. We have applied our method in DCE-MRIs of four different body parts: breast, brain, liver and prostate. The results show that we achieve average 89.5% success rate for 40 cases. The pharmacokinetic parameters computed from the automatic AIF are highly agreeable with those from a manually derived AIF (R2 = 0.89, P (T <=t) = 0.19) and a semiautomatic AIF (R2 = 0.98, P(T <=t) = 0.01).
Dynamic contrast-enhanced (DCE)-MRI is becoming an increasingly important tool for evaluating tumor vascularity and assessing the effectiveness of emerging antiangiogenic and antivascular agents. In chest and abdominal regions, however, respiratory motion can seriously degrade the achievable image quality in DCE-MRI studies. The purpose of this work is to develop a respiratory motion-compensated DCE-MRI technique that combines the self-gating properties of radial imaging with the reconstruction flexibility afforded by the golden-angle view-order strategy. Following radial data acquisition, the signal at k-space center is first used to determine the respiratory cycle, and consecutive views during the expiratory phase of each respiratory period (34-55 views, depending on the breathing rate) are grouped into individual segments. Residual intrasegment translation of lesion is subsequently compensated for by an autofocusing technique that optimizes image entropy, while intersegment translation (among different respiratory cycles) is corrected using 3D image correlation. The resulting motion-compensated, undersampled dynamic image series is then processed to reduce image streaking and to enhance the signal-to-noise ratio (SNR) prior to perfusion analysis, using either the k-space-weighted image contrast (KWIC) radial filtering technique or principal component analysis (PCA). The proposed data acquisition scheme also allows for high frame-rate arterial input function (AIF) sampling and free-breathing baseline T(1) mapping. The performance of the proposed radial DCE-MRI technique is evaluated in subjects with lung and liver lesions, and results demonstrate that excellent pixelwise perfusion maps can be obtained with the proposed methodology.
3D MR imaging of coronary arteries has the potential to provide both high resolution and high signal-to-noise ratio, but it is very susceptible to respiratory artifacts, especially respiratory blurring. Resolution loss caused by respiratory blurring in 3D coronary imaging is analyzed theoretically and verified experimentally. Under normal respiration, the width for any Gaussian point spread function is increased to a new value that is at least several millimeters (about 3-4 mm). In vivo studies were performed to compare respiratory pseudo-gated 3D acquisition with breath-hold 2D acquisition. On average, the overall quality of a pseudo-gated 3D image is worse than that of the corresponding breath-hold 2D image (P = 0.005). In most cases, respiratory blur caused coronary arteries in pseudo-gated 3D data to have lower resolution than in breath-hold 2D data.
An ECG-triggered magnetization-prepared segmented 3D fast gradient echo sequence was developed to perform pulmonary arterial MR angiography. A selective inversion recovery pulse was used in the magnetization preparation to suppress venous vasculature. A real-time gating technique based on navigator echoes was implemented to reduce respiration effects. Pencil-beam navigator echoes were acquired immediately before and after the readout train and processed in real-time to dynamically measure the diaphragm position, which was used to control data acquisition with an accept-or-reject-reacquire logic. In a study of 10 volunteers, a gated 3D acquisition with 28 slices required on average approximately 4 min of acquisition time, and six to seven segmental arteries related to the interlobar trunk of the pulmonary artery were depicted. The use of SIR pulse reduced venous signal by 99%. The gated acquisitions were superior to the ungated acquisitions (n = 10, P < 0.005). The real-time navigator gating technique is effective for reduction of respiration effects and thereby makes high resolution 3D MRA of the pulmonary arteries feasible.
We describe a standard set of quantity names and symbols related to the estimation of kinetic parameters from dynamic contrast-enhanced T(1)-weighted magnetic resonance imaging data, using diffusable agents such as gadopentetate dimeglumine (Gd-DTPA). These include a) the volume transfer constant K(trans) (min(-1)); b) the volume of extravascular extracellular space (EES) per unit volume of tissue v(e) (0 < v(e) < 1); and c) the flux rate constant between EES and plasma k(ep) (min(-1)). The rate constant is the ratio of the transfer constant to the EES (k(ep) = K(trans)/v(e)). Under flow-limited conditions K(trans) equals the blood plasma flow per unit volume of tissue; under permeability-limited conditions K(trans) equals the permeability surface area product per unit volume of tissue. We relate these quantities to previously published work from our groups; our future publications will refer to these standardized terms, and we propose that these be adopted as international standards.
Retrospective adaptive motion correction (AMC) was developed for reducing effects of residual respiration in real-time navigator-gated three-dimensional (3D) coronary magnetic resonance (MR) angiography. In both motion phantom and in vivo experiments, AMC improved image sharpness of coronary arteries. This navigator-based technique combining adaptive correction and real-time gating is potentially an efficient and effective motion reduction method for 3D coronary MR angiography.
Blood flow rate and velocity are important parameters for the study of vascular systems, and for the diagnosis, monitoring and evaluation of treatment of cerebro- and cardiovascular disease. For rapid imaging of cerebral and cardiac blood vessels, digital x-ray subtraction angiography has numerous advantages over other modalities. Roentgen-videodensitometric techniques measure blood flow and velocity from changes of contrast material density in x-ray angiograms. Many roentgen-videodensitometric flow measurement methods can also be applied to CT, MR and rotational angiography images. Hence, roentgen-videodensitometric blood flow and velocity measurement from digital x-ray angiograms represents an important research topic. This work contains a critical review and bibliography surveying current and old developments in the field. We present an extensive survey of English-language publications on the subject and a classification of published algorithms. We also present descriptions and critical reviews of these algorithms. The algorithms are reviewed with requirements imposed by neuro- and cardiovascular clinical environments in mind.
It has become increasingly important to quantitatively estimate tissue physiological parameters such as perfusion, capillary permeability, and the volume of extravascular-extracellular space (EES) using T(1)-weighted dynamic contrast-enhanced MRI (DCE-MRI). A linear equation was derived by integrating the differential equation describing the kinetic behavior of contrast agent (CA) in tissue, from which K(1) (rate constant for the transfer of CA from plasma to EES), k(2) (rate constant for the transfer from EES to plasma), and V(p) (plasma volume) can be easily obtained by the linear least-squares (LLSQ) method. The usefulness of this method was investigated by means of computer simulations, in comparison with the nonlinear least-squares (NLSQ) method. The new method calculated the above parameters faster than the NLSQ method by a factor of approximately 6, and estimated them more accurately than the NLSQ method at a signal-to-noise ratio (SNR) of < approximately 10. This method will be useful for generating functional images of K(1), k(2), and V(p) from DCE-MRI data.
Quantification of cerebral blood flow (CBF) using dynamic-susceptibility contrast (DSC) MRI relies on the deconvolution of the arterial input function (AIF). The AIF is commonly measured in a major artery (e.g., the middle cerebral artery), and the estimated function is used as a global AIF for the whole slice. However, the presence of bolus delay and dispersion between the artery and the tissue of interest can introduce significant errors in CBF quantification. While several methods have been introduced to minimize or eliminate the effects of bolus delay, the correction of bolus dispersion is more difficult to address because it requires a model for the vascular bed. This article summarizes how this dispersion effect can be incorporated into the model for CBF quantification, and discusses the magnitude of the errors introduced. Furthermore, alternative methods for correcting or minimizing the effects of bolus dispersion in the quantification of CBF are reviewed.
Rapid T(1)-weighted 3D spoiled gradient-echo (GRE) data sets were acquired in the abdomen of 23 cancer patients during a total of 113 separate visits to allow dynamic contrast-enhanced MRI (DCE-MRI) analysis of tumor microvasculature. The arterial input function (AIF) was measured in each patient at each visit using an automated AIF extraction method following a standardized bolus administration of gadodiamide. The AIFs for each patient were combined to obtain a mean AIF that is representative for any individual. The functional form of this general AIF may be useful for studies in which AIF measurements are not possible. Improvements in the reproducibility of DCE-MRI model parameters (K(trans), v(e), and v(p)) were observed when this new, high-temporal-resolution population AIF was used, indicating the potential for increased sensitivity to therapy-induced change.