Figure - available from: Scientific Reports
This content is subject to copyright. Terms and conditions apply.
BI-RADS 6 lesions in clinical cases (examples). Each tile of the figure presents the left or right breast of one clinical case with a diagnosed BI-RADS 6 lesion. Row 1 (A–E) shows images without the presence of artifacts. Row 2 (F–J) shows images that contain artifacts with no or moderate influence on the diagnostic assessment. Row 3 (K–O) shows images with artifacts that significantly impede the diagnostic evaluation. Artifacts in DWI often originate from multiple technical and/or patient-related sources that may be interdepend and thus it is not always possible to attribute one specific artifact source. The arrows mark regions of artifacts within the images with possible contributing factors related to distortion (e.g. visible in G), insufficient fat suppression (e.g. visible in I, L), related to remaining surface coil flare (e.g. also visible in I), and pulsation related signal drops in DWI (e.g. also visible in L).
Source publication
The objective of this IRB approved retrospective study was to apply deep learning to identify magnetic resonance imaging (MRI) artifacts on maximum intensity projections (MIP) of the breast, which were derived from diffusion weighted imaging (DWI) protocols. The dataset consisted of 1309 clinically indicated breast MRI examinations of 1158 individu...
Citations
... All cohort examinations (n=973) were previously included in studies focused on detecting artifacts in dynamic contrast-enhanced and DWI derived maximum-intensity projections (27,28). ...
Background: Virtual contrast-enhanced (vCE) imaging techniques are an emerging topic of research in breast MRI.
Purpose: To investigate how different combinations of T1-weighted (T1w), T2-weighted (T2w), and diffusion-weighted imaging (DWI) impact the performance of vCE breast MRI.
Materials and Methods: The IRB-approved, retrospective study included 1064 multiparametric breast MRI scans (age:52 ±12 years) obtained from 2017-2020 (single site, two 3T MRI). Eleven independent neural networks were trained to derive vCE images from varying input combinations of T1w, T2w, and multi-b-value DWI sequences (b-value=50-1500s/mm2). Three readers evaluated the vCE images with regards to qualitative scores of diagnostic image quality, image sharpness, satisfaction with contrast/signal-to-noise-ratio, and lesion/non-mass enhancement conspicuity. Quantitative metrics (SSIM, PSNR, NRMSE, and median symmetrical accuracy) were analyzed and statistically compared between the input combinations for the full breast volume and both enhancing and non-enhancing target findings.
Results: The independent test set consisted of 187 cases. The quantitative metrics significantly improved in target findings when multi-b-value DWI sequences were included during vCE training (p<.05). Non-significant effects (p>.05) were observed for the quantitative metrics on the full breast volume when comparing input combinations including T1w. Using T1w and DWI acquisitions during vCE training is necessary to achieve high satisfaction with contrast/SNR and good conspicuity of the enhancing findings. The input combination of T1w, T2w, and DWI sequences with three b-values showed the best qualitative performance.
Conclusion: vCE breast MRI performance is significantly influenced by input sequences. Quantitative metrics and visual quality of vCE images significantly benefit when a multi b-value DWI is added to morphologic T1w-/T2w-sequences as input for model training.
... All breast MRI examinations were conducted as part of the clinical routine. This cohort was part of previously published works by Liebert et al. [32] and Kapsner et al. [33,34], in which image quality assessment and artifact detection were investigated. The inclusion criteria were female patients with clinically indicated (e.g., preoperative or postoperative evaluations, assessment of multifocal disease or unclear findings, screening in cases of elevated breast cancer risk) breast MRI examination. ...
Several breast pathologies can affect the skin, and clinical pathways might differ significantly depending on the underlying diagnosis. This study investigates the feasibility of using diffusion-weighted imaging (DWI) to differentiate skin pathologies in breast MRIs. This retrospective study included 88 female patients who underwent diagnostic breast MRI (1.5 or 3T), including DWI. Skin areas were manually segmented, and the apparent diffusion coefficients (ADCs) were compared between different pathologies: inflammatory breast cancer (IBC; n = 5), benign skin inflammation (BSI; n = 11), Paget’s disease (PD; n = 3), and skin-involved breast cancer (SIBC; n = 11). Fifty-eight women had healthy skin (H; n = 58). The SIBC group had a significantly lower mean ADC than the BSI and IBC groups. These differences persisted for the first-order features of the ADC (mean, median, maximum, and minimum) only between the SIBC and BSI groups. The mean ADC did not differ significantly between the BSI and IBC groups. Quantitative DWI assessments demonstrated differences between various skin-affecting pathologies, but did not distinguish clearly between all of them. More extensive studies are needed to assess the utility of quantitative DWI in supplementing the diagnostic assessment of skin pathologies in breast imaging.
... CNNs are a type of deep learning model specifically designed to process image data by automatically and adaptively learning spatial hierarchies of features [10]. Through this process, CNNs can distinguish between normal and abnormal patterns in the image, enabling the identification and localization of artifacts, for example [6,[11][12][13][14][15]. However, such algorithms can only detect the artifacts after they have occurred, potentially making some acquisition unusable. ...
... Parts of the final patient population (n = 2265 and n = 1309) were included in previous studies by our group, dealing with the detection of artifacts in CE MIPs after the acquisition [6] and the detection of artifacts in MIPs of high b-value DWI [11]. Compared to these, this study focuses on the development of an algorithm focused on the prediction of an artifact before it occurs using just the pre-injection T1w acquisition of the CE MIP. ...
Objectives
To evaluate whether artifacts on contrast-enhanced (CE) breast MRI maximum intensity projections (MIPs) might already be forecast before gadolinium-based contrast agent (GBCA) administration during an ongoing examination by analyzing the unenhanced T1-weighted images acquired before the GBCA injection.
Materials and methods
This IRB-approved retrospective analysis consisted of n = 2884 breast CE MRI examinations after intravenous administration of GBCA, acquired with n = 4 different MRI devices at different field strengths (1.5 T/3 T) during clinical routine. CE-derived subtraction MIPs were used to conduct a multi-class multi-reader evaluation of the presence and severity of artifacts with three independent readers. An ensemble classifier (EC) of five DenseNet models was used to predict artifacts for the post-contrast subtraction MIPs, giving as the input source only the pre-contrast T1-weighted sequence. Thus, the acquisition directly preceded the GBCA injection. The area under ROC (AuROC) and diagnostics accuracy scores were used to assess the performance of the neural network in an independent holdout test set ( n = 285).
Results
After majority voting, potentially significant artifacts were detected in 53.6% ( n = 1521) of all breast MRI examinations (age 49.6 ± 12.6 years). In the holdout test set (mean age 49.7 ± 11.8 years), at a specificity level of 89%, the EC could forecast around one-third of artifacts (sensitivity 31%) before GBCA administration, with an AuROC = 0.66.
Conclusion
This study demonstrates the capability of a neural network to forecast the occurrence of artifacts on CE subtraction data before the GBCA administration. If confirmed in larger studies, this might enable a workflow-blended approach to prevent breast MRI artifacts by implementing in-scan personalized predictive algorithms.
Clinical relevance statement
Some artifacts in contrast-enhanced breast MRI maximum intensity projections might be predictable before gadolinium-based contrast agent injection using a neural network.
Key Points
• Potentially significant artifacts can be observed in a relevant proportion of breast MRI subtraction sequences after gadolinium-based contrast agent administration (GBCA).
• Forecasting the occurrence of such artifacts in subtraction maximum intensity projections before GBCA administration for individual patients was feasible at 89% specificity, which allowed correctly predicting one in three future artifacts.
• Further research is necessary to investigate the clinical value of such smart personalized imaging approaches.
In breast imaging, there is an unrelenting increase in the demand for breast imaging services, partly explained by continuous expanding imaging indications in breast diagnosis and treatment. As the human workforce providing these services is not growing at the same rate, the implementation of artificial intelligence (AI) in breast imaging has gained significant momentum to maximize workflow efficiency and increase productivity while concurrently improving diagnostic accuracy and patient outcomes. Thus far, the implementation of AI in breast imaging is at the most advanced stage with mammography and digital breast tomosynthesis techniques, followed by ultrasound, whereas the implementation of AI in breast magnetic resonance imaging (MRI) is not moving along as rapidly due to the complexity of MRI examinations and fewer available dataset. Nevertheless, there is persisting interest in AI‐enhanced breast MRI applications, even as the use of and indications of breast MRI continue to expand. This review presents an overview of the basic concepts of AI imaging analysis and subsequently reviews the use cases for AI‐enhanced MRI interpretation, that is, breast MRI triaging and lesion detection, lesion classification, prediction of treatment response, risk assessment, and image quality. Finally, it provides an outlook on the barriers and facilitators for the adoption of AI in breast MRI.
Level of Evidence
5
Technical Efficacy
Stage 6
The purpose of this feasibility study is to investigate if latent diffusion models (LDMs) are capable to generate contrast enhanced (CE) MRI-derived subtraction maximum intensity projections (MIPs) of the breast, which are conditioned by lesions. We trained an LDM with n = 2832 CE-MIPs of breast MRI examinations of n = 1966 patients (median age: 50 years) acquired between the years 2015 and 2020. The LDM was subsequently conditioned with n = 756 segmented lesions from n = 407 examinations, indicating their location and BI-RADS scores. By applying the LDM, synthetic images were generated from the segmentations of an independent validation dataset. Lesions, anatomical correctness, and realistic impression of synthetic and real MIP images were further assessed in a multi-rater study with five independent raters, each evaluating n = 204 MIPs (50% real/50% synthetic images). The detection of synthetic MIPs by the raters was akin to random guessing with an AUC of 0.58. Interrater reliability of the lesion assessment was high both for real (Kendall’s W = 0.77) and synthetic images (W = 0.85). A higher AUC was observed for the detection of suspicious lesions (BI-RADS ≥\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge $$\end{document} 4) in synthetic MIPs (0.88 vs. 0.77; p = 0.051). Our results show that LDMs can generate lesion-conditioned MRI-derived CE subtraction MIPs of the breast, however, they also indicate that the LDM tended to generate rather typical or ‘textbook representations’ of lesions.
