Bibo Shi1, Lars J Grimm2, Maciej A Mazurowski2, Jay A Baker2, Jeffrey R Marks3, Lorraine M King3, Carlo C Maley4, E Shelley Hwang3, Joseph Y Lo2. 1. Carl E. Ravin Advanced Imaging Laboratories, Department of Radiology, Duke University School of Medicine, Durham, North Carolina. Electronic address: bibo.shi@duke.edu. 2. Carl E. Ravin Advanced Imaging Laboratories, Department of Radiology, Duke University School of Medicine, Durham, North Carolina. 3. Department of Surgery, Duke University School of Medicine, Durham, North Carolina. 4. Biodesign Center for Personalized Diagnostics and School of Life Sciences, Arizona State University, Tempe, Arizona; Centre for Evolution and Cancer, Institute of Cancer Research, London, United Kingdom.
Abstract
PURPOSE: The aim of this study was to determine whether deep features extracted from digital mammograms using a pretrained deep convolutional neural network are prognostic of occult invasive disease for patients with ductal carcinoma in situ (DCIS) on core needle biopsy. METHODS: In this retrospective study, digital mammographic magnification views were collected for 99 subjects with DCIS at biopsy, 25 of which were subsequently upstaged to invasive cancer. A deep convolutional neural network model that was pretrained on nonmedical images (eg, animals, plants, instruments) was used as the feature extractor. Through a statistical pooling strategy, deep features were extracted at different levels of convolutional layers from the lesion areas, without sacrificing the original resolution or distorting the underlying topology. A multivariate classifier was then trained to predict which tumors contain occult invasive disease. This was compared with the performance of traditional "handcrafted" computer vision (CV) features previously developed specifically to assess mammographic calcifications. The generalization performance was assessed using Monte Carlo cross-validation and receiver operating characteristic curve analysis. RESULTS: Deep features were able to distinguish DCIS with occult invasion from pure DCIS, with an area under the receiver operating characteristic curve of 0.70 (95% confidence interval, 0.68-0.73). This performance was comparable with the handcrafted CV features (area under the curve = 0.68; 95% confidence interval, 0.66-0.71) that were designed with prior domain knowledge. CONCLUSIONS: Despite being pretrained on only nonmedical images, the deep features extracted from digital mammograms demonstrated comparable performance with handcrafted CV features for the challenging task of predicting DCIS upstaging.
PURPOSE: The aim of this study was to determine whether deep features extracted from digital mammograms using a pretrained deep convolutional neural network are prognostic of occult invasive disease for patients with ductal carcinoma in situ (DCIS) on core needle biopsy. METHODS: In this retrospective study, digital mammographic magnification views were collected for 99 subjects with DCIS at biopsy, 25 of which were subsequently upstaged to invasive cancer. A deep convolutional neural network model that was pretrained on nonmedical images (eg, animals, plants, instruments) was used as the feature extractor. Through a statistical pooling strategy, deep features were extracted at different levels of convolutional layers from the lesion areas, without sacrificing the original resolution or distorting the underlying topology. A multivariate classifier was then trained to predict which tumors contain occult invasive disease. This was compared with the performance of traditional "handcrafted" computer vision (CV) features previously developed specifically to assess mammographic calcifications. The generalization performance was assessed using Monte Carlo cross-validation and receiver operating characteristic curve analysis. RESULTS: Deep features were able to distinguish DCIS with occult invasion from pure DCIS, with an area under the receiver operating characteristic curve of 0.70 (95% confidence interval, 0.68-0.73). This performance was comparable with the handcrafted CV features (area under the curve = 0.68; 95% confidence interval, 0.66-0.71) that were designed with prior domain knowledge. CONCLUSIONS: Despite being pretrained on only nonmedical images, the deep features extracted from digital mammograms demonstrated comparable performance with handcrafted CV features for the challenging task of predicting DCIS upstaging.
Authors: Mary F Dillon; Enda W McDermott; Cecily M Quinn; Ann O'Doherty; Niall O'Higgins; Arnold D K Hill Journal: J Surg Oncol Date: 2006-06-01 Impact factor: 3.454
Authors: Rahul Paul; Samuel H Hawkins; Yoganand Balagurunathan; Matthew B Schabath; Robert J Gillies; Lawrence O Hall; Dmitry B Goldgof Journal: Tomography Date: 2016-12
Authors: Rui Hou; Maciej A Mazurowski; Lars J Grimm; Jeffrey R Marks; Lorraine M King; Carlo C Maley; Eun-Sil Shelley Hwang; Joseph Y Lo Journal: IEEE Trans Biomed Eng Date: 2019-09-09 Impact factor: 4.538
Authors: Simukayi Mutasa; Peter Chang; Eduardo P Van Sant; John Nemer; Michael Liu; Jenika Karcich; Gita Patel; Sachin Jambawalikar; Richard Ha Journal: Acad Radiol Date: 2019-09-14 Impact factor: 3.173