In the last two decades, Computer Aided Detection (CAD) systems were developed to help radiologists analyse screening mammograms, however benefits of current CAD technologies appear to be contradictory, therefore they should be improved to be ultimately considered useful. Since 2012, deep convolutional neural networks (CNN) have been a tremendous success in image recognition, reaching human performance. These methods have greatly surpassed the traditional approaches, which are similar to currently used CAD solutions. Deep CNN-s have the potential to revolutionize medical image analysis. We propose a CAD system based on one of the most successful object detection frameworks, Faster R-CNN. The system detects and classifies malignant or benign lesions on a mammogram without any human intervention. The proposed method sets the state of the art classification performance on the public INbreast database, AUC = 0.95. The approach described here has achieved 2nd place in the Digital Mammography DREAM Challenge with AUC = 0.85. When used as a detector, the system reaches high sensitivity with very few false positive marks per image on the INbreast dataset. Source code, the trained model and an OsiriX plugin are published online at https://github.com/riblidezso/frcnn_cad .

/pdf/detecting-and-classifying-lesions-in-mammograms-with-deep-41p9rqz0mh.pdf

Detecting and classifying lesions in mammograms with Deep Learning

Published research results are difficult to replicate due to the lack of a standard evaluation data set in the area of decision support systems in mammography; most computer-aided diagnosis (CADx) and detection (CADe) algorithms for breast cancer in mammography are evaluated on private data sets or on unspecified subsets of public databases. This causes an inability to directly compare the performance of methods or to replicate prior results. We seek to resolve this substantial challenge by releasing an updated and standardized version of the Digital Database for Screening Mammography (DDSM) for evaluation of future CADx and CADe systems (sometimes referred to generally as CAD) research in mammography. Our data set, the CBIS-DDSM (Curated Breast Imaging Subset of DDSM), includes decompressed images, data selection and curation by trained mammographers, updated mass segmentation and bounding boxes, and pathologic diagnosis for training data, formatted similarly to modern computer vision data sets. The data set contains 753 calcification cases and 891 mass cases, providing a data-set size capable of analyzing decision support systems in mammography.

/pdf/a-curated-mammography-data-set-for-use-in-computer-aided-2g8dyfxrop.pdf

A curated mammography data set for use in computer-aided detection and diagnosis research.

In the last two decades Computer Aided Diagnostics (CAD) systems were developed to help radiologists analyze screening mammograms. The benefits of current CAD technologies appear to be contradictory and they should be improved to be ultimately considered useful. Since 2012 deep convolutional neural networks (CNN) have been a tremendous success in image recognition, reaching human performance. These methods have greatly surpassed the traditional approaches, which are similar to currently used CAD solutions. Deep CNN-s have the potential to revolutionize medical image analysis. We propose a CAD system based on one of the most successful object detection frameworks, Faster R-CNN. The system detects and classifies malignant or benign lesions on a mammogram without any human intervention. The proposed method sets the state of the art classification performance on the public INbreast database, AUC = 0.95 . The approach described here has achieved the 2nd place in the Digital Mammography DREAM Challenge with AUC = 0.85 . When used as a detector, the system reaches high sensitivity with very few false positive marks per image on the INbreast dataset. Source code, the trained model and an OsiriX plugin are availaible online at this https URL .

Machine learning techniques for breast cancer computer aided diagnosis using different image modalities: A systematic review.

Breast cancer is a common and fatal disease among women worldwide. Therefore, the early and precise diagnosis of breast cancer plays a pivotal role to improve the prognosis of patients with this disease. Several studies have developed automated techniques using different medical imaging modalities to predict breast cancer development. However, few review studies are available to recapitulate the existing literature on breast cancer classification. These studies provide an overview of the classification, segmentation, or grading of many cancer types, including breast cancer, by using traditional machine learning approaches through hand-engineered features. This review focuses on breast cancer classification by using medical imaging multimodalities through state-of-the-art artificial deep neural network approaches. It is anticipated to maximize the procedural decision analysis in five aspects, such as types of imaging modalities, datasets and their categories, pre-processing techniques, types of deep neural network, and performance metrics used for breast cancer classification. Forty-nine journal and conference publications from eight academic repositories were methodically selected and carefully reviewed from the perspective of the five aforementioned aspects. In addition, this study provided quantitative, qualitative, and critical analyses of the five aspects. This review showed that mammograms and histopathologic images were mostly used to classify breast cancer. Moreover, about 55% of the selected studies used public datasets, and the remaining used exclusive datasets. Several studies employed augmentation, scaling, and image normalization pre-processing techniques to minimize inconsistencies in breast cancer images. Several types of shallow and deep neural network architecture were employed to classify breast cancer using images. The convolutional neural network was utilized frequently to construct an effective breast cancer classification model. Some of the selected studies employed a pre-trained network or developed new deep neural networks to classify breast cancer. Most of the selected studies used accuracy and area-under-the-curve metrics followed by sensitivity, precision, and F-measure metrics to evaluate the performance of the developed breast cancer classification models. Finally, this review presented 10 open research challenges for future scholars who are interested to develop breast cancer classification models through various imaging modalities. This review could serve as a valuable resource for beginners on medical image classification and for advanced scientists focusing on deep learning-based breast cancer classification through different medical imaging modalities.

Deep learning-based breast cancer classification through medical imaging modalities: state of the art and research challenges

To evaluate the frequency of diagnosis of high-risk lesions at MRI-guided vacuum-assisted breast biopsy (MRgVABB) and to determine whether underestimation may be predicted. Retrospective review of the medical records of 161 patients who underwent MRgVABB was performed. The underestimation rate was defined as an upgrade of a high-risk lesion at MRgVABB to malignancy at surgery. Clinical data, MRI features of the biopsied lesions, and histological diagnosis of cases with and those without underestimation were compared. Of 161 MRgVABB, histology revealed 31 (19%) high-risk lesions. Of 26 excised high-risk lesions, 13 (50%) were upgraded to malignancy. The underestimation rates of lobular neoplasia, atypical apocrine metaplasia, atypical ductal hyperplasia, and flat epithelial atypia were 50% (4/8), 100% (5/5), 50% (3/6) and 50% (1/2) respectively. There was no underestimation in the cases of benign papilloma without atypia (0/3), and radial scar (0/2). No statistically significant differences (p > 0.1) between the cases with and those without underestimation were seen in patient age, indications for breast MRI, size of lesion on MRI, morphological and kinetic features of biopsied lesions. Imaging and clinical features cannot be used reliably to predict underestimation at MRgVABB. All high-risk lesions diagnosed at MRgVABB require surgical excision.

/pdf/high-risk-lesions-diagnosed-at-mri-guided-vacuum-assisted-4mchbvlumi.pdf

Arifa Sadaf

Papers

High-risk lesions diagnosed at MRI-guided vacuum-assisted breast biopsy: can underestimation be predicted?

Performance of computer-aided detection applied to full-field digital mammography in detection of breast cancers