Rory Wilding1, Vivek M Sheraton1, Lysabella Soto2,3, Niketa Chotai4, Ern Yu Tan5. 1. Onkolyze Pte. Ltd., Singapore, Singapore. 2. Research Department, RESOMED, Maracaibo, Venezuela. 3. Postgraduate Division Studies of Radiology, Medicine School of Zulia's University, Maracaibo, Venezuela. 4. Department of Radiology, RadLink Diagnostic Imaging Center, Singapore, Singapore. 5. Department of General Surgery, Tan Tock Seng Hospital, Singapore, Singapore. ern_yu_tan@ttsh.com.sg.
Abstract
PURPOSE: Automatic classification and segmentation of tumors in breast ultrasound images enables better diagnosis and planning treatment strategies for breast cancer patients. METHODS: We collected 953 breast ultrasound images from two open-source datasets and classified them with help of an expert radiologist according to BI-RADS criteria. The data was split into normal, benign and malignant classes. We then used machine learning to develop classification and segmentation algorithms. RESULTS: We found 3.92% of the images across the open-source datasets had erroneous classifications. Post-radiologist intervention, three algorithms were developed based on the classification categories. Classification algorithms distinguished images with healthy breast tissue from those with abnormal tissue with 96% accuracy, and distinguished benign from malignant images with 85% accuracy. Both algorithms generated robust F1 and AUROC metrics. Finally, the masses within images were segmented with an 80.31% DICE score. CONCLUSIONS: Our work illustrates the potential of deep learning algorithms to improve the accuracy of breast ultrasound assessments and to facilitate automated assessments.
PURPOSE: Automatic classification and segmentation of tumors in breast ultrasound images enables better diagnosis and planning treatment strategies for breast cancer patients. METHODS: We collected 953 breast ultrasound images from two open-source datasets and classified them with help of an expert radiologist according to BI-RADS criteria. The data was split into normal, benign and malignant classes. We then used machine learning to develop classification and segmentation algorithms. RESULTS: We found 3.92% of the images across the open-source datasets had erroneous classifications. Post-radiologist intervention, three algorithms were developed based on the classification categories. Classification algorithms distinguished images with healthy breast tissue from those with abnormal tissue with 96% accuracy, and distinguished benign from malignant images with 85% accuracy. Both algorithms generated robust F1 and AUROC metrics. Finally, the masses within images were segmented with an 80.31% DICE score. CONCLUSIONS: Our work illustrates the potential of deep learning algorithms to improve the accuracy of breast ultrasound assessments and to facilitate automated assessments.
Authors: Moi Hoon Yap; Gerard Pons; Joan Marti; Sergi Ganau; Melcior Sentis; Reyer Zwiggelaar; Adrian K Davison; Robert Marti; Gerard Pons; Joan Marti; Sergi Ganau; Melcior Sentis; Reyer Zwiggelaar; Adrian K Davison; Robert Marti Journal: IEEE J Biomed Health Inform Date: 2017-08-07 Impact factor: 5.772
Authors: Alexander Ciritsis; Cristina Rossi; Matthias Eberhard; Magda Marcon; Anton S Becker; Andreas Boss Journal: Eur Radiol Date: 2019-03-29 Impact factor: 5.315
Authors: Anna Maier; Joerg Heil; Anna Lauer; Aba Harcos; Benedikt Schaefgen; Alexandra von Au; Julia Spratte; Fabian Riedel; Geraldine Rauch; André Hennigs; Christoph Domschke; Sarah Schott; Joachim Rom; Florian Schuetz; Christof Sohn; Michael Golatta Journal: Arch Gynecol Obstet Date: 2017-07-26 Impact factor: 2.344
Authors: Wendie A Berg; Jeffrey D Blume; Jean B Cormack; Ellen B Mendelson; Daniel Lehrer; Marcela Böhm-Vélez; Etta D Pisano; Roberta A Jong; W Phil Evans; Marilyn J Morton; Mary C Mahoney; Linda Hovanessian Larsen; Richard G Barr; Dione M Farria; Helga S Marques; Karan Boparai Journal: JAMA Date: 2008-05-14 Impact factor: 56.272