Fig 2 - uploaded by Wenjing Li
Content may be subject to copyright.
The concept of color calibration: mapping the raw color space of different instruments into a standard color space.
Source publication
Colposcopy is a primary diagnostic method used to detect cancer and precancerous lesions of the uterine cervix. During the examination, the metaplastic and abnormal tissues exhibit different degrees of whiteness (acetowhitening effect) after applying a 3%-5% acetic acid solution. Colposcopists evaluate the color and density of the acetowhite tissue...
Context in source publication
Context 1
... image often appear different from what is perceived by eye. The goal of image calibration is that the colors should appear to be identical, independent of camera/camera settings and light source used. This can be achieved by mapping the color appearance of the images taken with different instruments into a standard color space, as illustrated in Fig. 2. The entire calibration procedure proposed for the colposcopic image calibration is shown in Fig. 3. Both exam data and calibration data are acquired at the clinical sites using the same instrument. Calibration data includes images of a gray target for gray balance and a color target for color calibration (see Fig. 4). The image of the ...
Similar publications
Colposcopy is a primary diagnostic method used to detect cancer and precancerous lesions of the uterine cervix. During the examination, the metaplastic and abnormal tissues exhibit different degrees of whiteness (acetowhitening effect) after applying a 3%-5% acetic acid solution. Colposcopists evaluate the color and density of the acetowhite tissue...
Citations
... However, the performance of the aforementioned studies has been tested by only performing the visual assessment of their results, no objective assessment has been performed and the visual result seems similar for all the methods. There are certain other contributions, suggested by Li et al. [19] and Rouhbakhsh et al. [20] for improving the quality of images by performing contrast and color normalization. The results of different techniques suggested by the distinct authors are illustrated in Fig. 3. ...
Cervical cancer is one among the trivial forms of cancer that counts for 6.6% of all females cancers with an estimated 570,000 new cases in 2018.
The mortality rate due to cervical cancer is approximately 90% in low or middle income countries due to lack of suitable pre-screening procedures and experienced medical staff. Colposcopy images or cervigrams, are the images that capture the cervical region, are considered as the gold standard by the medical experts for the identification and evaluation of cervical cancer. The visual assessment of cervigrams for recognizing cancer suffers from high inter- or intra-variations especially among less or unskilled medical experts. However, this method is dependent on colposcopists’ observation and it is more time consuming, tedious and laborious task which calls for development of computer-aided method for diagnosis of cervical cancer. With the technological advancements, deep learning has been commonly employed for providing automated solutions for disease diagnosis due to its self-learning capability. This paper presents a deep-learning-based method for cervix cancer classification using colposcopy images. The architecture of the proposed method namely, ColpoNet, has been motivated by the DenseNet model because it is computationally more efficient as compared to other models. Further, the method has been tested and validated on the dataset released by the National Cancer Institute and it has been compared with other deep-learning models namely AlexNet, VGG16, ResNet50, LeNet and GoogleNet to check scope of its applicability. The experimental analysis revealed that ColpoNet achieved an accuracy of 81.353% and shows the highest performance rate as compared to other state-of-the-art deep techniques. Such classification system can be deployed in clinics to enhance the early detection of cervical cancer in less developed countries.
... In addition to the aforementioned competition on the analysis of vulnerable population, Intel and MobileODT organized a competition for the automatic analysis of digital colposcopies in 2017 [53]. This increasing interest has resulted in a stable community with well identified problems that range, from the quality assessment [23] and enhancement [39], [70] of digital colposcopies, to the segmentation of the anatomical parts of the cervix [14], [71], to the final diagnosis [95], [116], [117]. While the vast majority of databases that were used in the development of these papers are closed, as a result of these competitions, new public databases of considerable size and with new challenges were released [46], [53]. ...
... Several works have been proposed in the area of quality enhancement of colposcopic images [13], [14], [39], [60], [64], [68], [70], [94], most of them focusing on the removal of specular reflections (SR) [13], [14], [39], [60], [64], [68]. The remaining works, proposed by Li et al. [70] and Rouhbakhsh et al. [94] focused on the enhancement of images by means of color and contrast normalization. ...
... Several works have been proposed in the area of quality enhancement of colposcopic images [13], [14], [39], [60], [64], [68], [70], [94], most of them focusing on the removal of specular reflections (SR) [13], [14], [39], [60], [64], [68]. The remaining works, proposed by Li et al. [70] and Rouhbakhsh et al. [94] focused on the enhancement of images by means of color and contrast normalization. It is relevant to highlight that image enhancement can be done with two goals in mind, which may lead to different techniques and evaluation settings. ...
Cervical cancer remains a significant cause of mortality in low-income countries. However, it can often be cured by removing the affected tissues when detected in early stages. Thereby, it is relevant to provide universal and efficient access to cervical screening programs, being digital colposcopy an inexpensive technique with high potential of scalability. The development of Computer-aided Diagnosis systems for the automated processing of digital colposcopies has gained the attention of the computer vision and machine learning communities in the last decade, giving origin to a wide diversity of tasks and computational solutions. However, there is a lack of a unified framework to discuss the main tasks and to assess their performance. Thus, in this work, we studied the core research lines surrounding the automated analysis of digital colposcopies and built a topology of problems and techniques, including their key properties, advantages, and limitations. Also, we discussed the open challenges in the area and released a database that serves as a common basis to evaluate such systems.
... Several color calibration algorithms are proposed for various tasks in different device-independent color spaces. The most commonly used methods include the Polynomial-Based Regression, the Neural Network mapping algorithms, the Support Vector Regression, and the Ridge Regression [18][19][20][21][22][23][24][25][26][27][28][29]. Moreover, these color calibration methods have been applied into the general imaging devices [16,[18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33], such as digital cameras [16,[18][19][20][21][25][26][27], digital colposcopy [22], scanners [23,24,30], printers and cathode ray tube/liquid crystal display (CRT/LCD) monitors [23,28], and tristimulus colorimeter [31][32][33]. ...
... The most commonly used methods include the Polynomial-Based Regression, the Neural Network mapping algorithms, the Support Vector Regression, and the Ridge Regression [18][19][20][21][22][23][24][25][26][27][28][29]. Moreover, these color calibration methods have been applied into the general imaging devices [16,[18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33], such as digital cameras [16,[18][19][20][21][25][26][27], digital colposcopy [22], scanners [23,24,30], printers and cathode ray tube/liquid crystal display (CRT/LCD) monitors [23,28], and tristimulus colorimeter [31][32][33]. Most related researches put emphasis on the comparison of color calibration methods for various tasks. ...
... The most commonly used methods include the Polynomial-Based Regression, the Neural Network mapping algorithms, the Support Vector Regression, and the Ridge Regression [18][19][20][21][22][23][24][25][26][27][28][29]. Moreover, these color calibration methods have been applied into the general imaging devices [16,[18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33], such as digital cameras [16,[18][19][20][21][25][26][27], digital colposcopy [22], scanners [23,24,30], printers and cathode ray tube/liquid crystal display (CRT/LCD) monitors [23,28], and tristimulus colorimeter [31][32][33]. Most related researches put emphasis on the comparison of color calibration methods for various tasks. ...
Color measurement by the colorized vision system is a superior method to achieve the evaluation of color objectively and continuously. However, the accuracy of color measurement is influenced by the spectral responses of digital sensor and the spectral mismatch of illumination. In this paper, two-color vision system illuminated by digital sensor and LED array, respectively, is presented. The Polynomial-Based Regression method is applied to solve the problem of color calibration in the sRGB and C I E L ⁎ a ⁎ b ⁎ color spaces. By mapping the tristimulus values from RGB to sRGB color space, color difference between the estimated values and the reference values is less than 3 Δ E . Additionally, the mapping matrix Φ R G B → s R G B has proved a better performance in reducing the color difference, and it is introduced subsequently into the colorized vision system proposed for a better color measurement. Necessarily, the printed matter of clothes and the colored ceramic tile are chosen as the application experiment samples of our colorized vision system. As shown in the experimental data, the average color difference of images is less than 6 Δ E . It indicates that a better performance of color measurement is obtained via the colorized vision system proposed.
... There is no currently internationally recognized standard methodology for assessing imaging performance of digital colposcopes [25]. Prior groups have reported a battery of targets and tests for performance assessment and/or digital colposcope calibration [26][27][28]. We first set out to confirm the reported imaging specifications of these different systems including: resolving power (micron), working distance (mm), and diagonal field of view (mm), at a range of possible magnifications, if the system was capable of multiple magnifications. ...
... The opacity and color of potential lesions are important characteristics often used to stage cervical pre-cancers and accurate color reproduction is needed to prevent misclassification by the clinician. Accuracy in white balance is important as low-grade lesions are often characterized as having snow to bright white coloration when compared to high-grade lesions that are oyster gray in color [26,34,35]. In white field imaging, red color reproduction and in green field imaging green color reproduction is important for detection underlying vasculature [34]. ...
... The use of a narrower band of illumination helps enhances the vasculature Novel Low Cost POCkeT Digital Colposcope contrast when compared to broader illumination [37]. Non-uniform light distribution is a major hurdle in image processing for lesion margin and vasculature extraction from digital cervical images [26,38,39]. Thus, beam characterization of each systems illumination strategy provide was examined in the case of white field illumination with contour plots of luminance overlaid on images on a diffuse target (Fig 5C to 5G). ...
Current guidelines by WHO for cervical cancer screening in low- and middle-income countries involves visual inspection with acetic acid (VIA) of the cervix, followed by treatment during the same visit or a subsequent visit with cryotherapy if a suspicious lesion is found. Implementation of these guidelines is hampered by a lack of: trained health workers, reliable technology, and access to screening facilities. A low cost ultra-portable Point of Care Tampon based digital colposcope (POCkeT Colposcope) for use at the community level setting, which has the unique form factor of a tampon, can be inserted into the vagina to capture images of the cervix, which are on par with that of a state of the art colposcope, at a fraction of the cost. A repository of images to be compiled that can be used to empower front line workers to become more effective through virtual dynamic training. By task shifting to the community setting, this technology could potentially provide significantly greater cervical screening access to where the most vulnerable women live. The POCkeT Colposcope's concentric LED ring provides comparable white and green field illumination at a fraction of the electrical power required in commercial colposcopes. Evaluation with standard optical imaging targets to assess the POCkeT Colposcope against the state of the art digital colposcope and other VIAM technologies.
Our POCkeT Colposcope has comparable resolving power, color reproduction accuracy, minimal lens distortion, and illumination when compared to commercially available colposcopes. In vitro and pilot in vivo imaging results are promising with our POCkeT Colposcope capturing comparable quality images to commercial systems.
The POCkeT Colposcope is capable of capturing images suitable for cervical lesion analysis. Our portable low cost system could potentially increase access to cervical cancer screening in limited resource settings through task shifting to community health workers.
... Various types of relationships have been already implemented in the literature, such as multidimensional lookup tables with interpolation678 , least-squares polynomial regressions with various polynomial orders, advanced regression modeling [9], neural networks [10,11] and also human-observation-based models [12]. These color calibration methods have been implemented successfully for various imaging systems: computer vision systems for meat quality evaluation [9,13], dermatoscopic imaging systems141516, colposcopes [17] , commercial digital cameras for various applications such as dentistry [18] or color advice for home décor [19], displays [7,12], or printers [8,11]. Generally color calibration algorithms are not very robust to changes in the illumination conditions, since the color of objects depends on the illumination source. ...
This paper aims at showing that performing color calibration of an RGB camera can be achieved even in the case where the optical system before the camera introduces strong color distortion. In the present case, the optical system is a microscope containing a halogen lamp, with a nonuniform irradiance on the viewed surface. The calibration method proposed in this work is based on an existing method, but it is preceded by a three-step preprocessing of the RGB images aiming at extracting relevant color information from the strongly distorted images, taking especially into account the nonuniform irradiance map and the perturbing texture due to the surface topology of the standard color calibration charts when observed at micrometric scale. The proposed color calibration process consists first in computing the average color of the color-chart patches viewed under the microscope; then computing white balance, gamma correction, and saturation enhancement; and finally applying a third-order polynomial regression color calibration transform. Despite the nonusual conditions for color calibration, fairly good performance is achieved from a 48 patch Lambertian color chart, since an average CIE-94 color difference on the color-chart colors lower than 2.5 units is obtained.
... We generate a color correction matrix that transforms the values extracted from the image of the color target to standard values in the preferred color space. More detailed procedures are described in [31]. ...
This paper presents a domain-specific automated image analysis framework for the detection of pre-cancerous and cancerous lesions of the uterine cervix. Our proposed framework departs from previous methods in that we include domain-specific diagnostic features in a probabilistic manner using conditional random fields. Likewise, we provide a novel window-based performance assessment scheme for 2D image analysis which addresses the intrinsic problem of image misalignment. Image regions corresponding to different tissue types are indentified for the extraction of domain-specific anatomical features. The unique optical properties of each tissue type and the diagnostic relationships between neighboring regions are incorporated in the proposed conditional random field model. The validity of our method is examined using clinical data from 48 patients, and its diagnostic potential is demonstrated by a performance comparison with expert colposcopy annotations, using histopathology as the ground truth. The proposed automated diagnostic approach can support or potentially replace conventional colposcopy, allow tissue specimen sampling to be performed in a more objective manner, and lower the number of cervical cancer cases in developing countries by providing a cost effective screening solution in low-resource settings.
... Furthermore, a device with a quality control regimen and an automated image analysis would possibly serve as a replacement to the current diagnostic technologies. A variety of studies have demonstrated the importance of an image quality measure by investigating the functionality of imaging equipment based on its resolution and contrast power, but only a few have deployed it in a clinical setting789. Several groups are developing and testing cervical cancer screening and diagnostic systems, but are deficient of a wellorganized quality control program that would rigorously and frequently test the performance and reliability of both hardware and software components in real time [9]. ...
... Lualdi et al. have tested their quality assurance protocol using tissue-like phantoms on a series of 30 Spectro- Shade® instruments and reported the results of inter-and intrainstrument variability [9]. Additionally, Wenjing Li et al. have implemented a color calibration unit in a computer aided diagnosis system for colposcopy that would correct for the nonuniform nature of the light source and also calibrate the images to its true color space [7,8]. The same team has designed an image quality assessment program for identifying incorrect instrumentation settings and for testing the quality of only reflectance images. ...
The diagnostic ability of algorithms developed for the Multispectral Digital Colposcope (MDC) is highly dependent on the quality of the image. The field of objective medical image quality analysis has great potential but has not been well exploited. Various researchers have reported different measures of image quality but with an existence of a reference image. The quality of an image can be attributed to several sources of errors, a few of which would be inclusion of presence of extraneous components, improper illumination, or an image out of focus. This can be due to motion artifact or the region of interest out of the focal plane.
With spectroscopic measurements, assessment of data quality has been used by our group in the past to avoid hardware errors at the time of acquisition. We are currently developing algorithms that will help identify hardware and acquisition errors to the clinician in under a few seconds.
Minimizing these errors not only provides quality images for a diagnostic algorithm, but reduces the necessity for complex and time intensive post-processing software for enhancing the images.
We propose a no reference image quality system specifically designed for MDC that can be modified to similar spectroscopic imaging applications.
... Besides, the non-uniform illumination of the light source is also barrier for image processing algorithms. We apply an image calibration procedure based on gray and color target balancing to calibrate the images in terms of illumination and color 6 . An example of a calibrated image is displayed in Fig. 1 ...
Cervical Cancer is the second most common cancer among women worldwide and the leading cause of cancer mortality of women in developing countries. If detected early and treated adequately, cervical cancer can be virtually prevented. Cervical precursor lesions and invasive cancer exhibit certain morphologic features that can be identified during a visual inspection exam. Digital imaging technologies allow us to assist the physician with a Computer-Aided Diagnosis (CAD) system. In colposcopy, epithelium that turns white after application of acetic acid is called acetowhite epithelium. Acetowhite epithelium is one of the major diagnostic features observed in detecting cancer and pre-cancerous regions. Automatic extraction of acetowhite regions from cervical images has been a challenging task due to specular reflection, various illumination conditions, and most importantly, large intra-patient variation. This paper presents a multi-step acetowhite region detection system to analyze the acetowhite lesions in cervical images automatically. First, the system calibrates the color of the cervical images to be independent of screening devices. Second, the anatomy of the uterine cervix is analyzed in terms of cervix region, external os region, columnar region, and squamous region. Third, the squamous region is further analyzed and subregions based on three levels of acetowhite are identified. The extracted acetowhite regions are accompanied by color scores to indicate the different levels of acetowhite. The system has been evaluated by 40 human subjects' data and demonstrates high correlation with experts' annotations.
Colposcopy plays a vital role in detecting cervical cancer. Artificial intelligence-based methods have been implemented in the literature for the classification of colposcopy images. However, there is a need for a more effective method that can accurately classify cervigrams. In this paper, ColpoClassifier, a hybrid framework for the classification of cervigrams, is proposed, which consists of feature extraction followed by classification. This paper uses a Gray-level co-occurrence matrix (GLCM), a Gray-level run length matrix (GLRLM), and a histogram of gradients (HOG) for feature extraction. These features are combined to form a feature fusion vector of the form GLCM + GLRLM + HOG. The different machine learning classifiers are used for classification by using individual feature vectors as well as feature fusion vectors. The dataset used in this paper is compiled by downloading images from the WHO website. Two variants of this dataset are created, Dataset-I contains images of the aceto-whitening effect, green filter, iodine application, and raw cervigram while Dataset-II only contains images of the aceto-whitening effect. This paper presents the classification performance on all kinds of images with the individual as well as hybrid feature fusion vector and concludes that hybrid feature fusion vectors on aceto-whitening images have given the best results.
