Yoichiro Yamamoto , 1 , 2 , Toyonori Tsuzuki 3 , Jun Akatsuka 1 , 4 , Masao Ueki 5 , Hiromu Morikawa 1 , Yasushi Numata 1 , Taishi Takahara 3 , Takuji Tsuyuki 3 , Kotaro Tsutsumi 1 , Ryuto Nakazawa 6 , Akira Shimizu 7 , Ichiro Maeda 1 , 8 , Shinichi Tsuchiya 9 , Hiroyuki Kanno 2 , Yukihiro Kondo 4 , Manabu Fukumoto 1 , 10 , Gen Tamiya 5 , 11 , Naonori Ueda 12 , Go Kimura , 4
18 December 2019
Deep learning algorithms have been successfully used in medical image classification. In the next stage, the technology of acquiring explainable knowledge from medical images is highly desired. Here we show that deep learning algorithm enables automated acquisition of explainable features from diagnostic annotation-free histopathology images. We compare the prediction accuracy of prostate cancer recurrence using our algorithm-generated features with that of diagnosis by expert pathologists using established criteria on 13,188 whole-mount pathology images consisting of over 86 billion image patches. Our method not only reveals findings established by humans but also features that have not been recognized, showing higher accuracy than human in prognostic prediction. Combining both our algorithm-generated features and human-established criteria predicts the recurrence more accurately than using either method alone. We confirm robustness of our method using external validation datasets including 2276 pathology images. This study opens up fields of machine learning analysis for discovering uncharted knowledge.
Technologies for acquiring explainable features from medical images need further development. Here, the authors report a deep learning based automated acquisition of explainable features from pathology images, and show a higher accuracy of their method as compared to pathologist based diagnosis of prostate cancer recurrence.