Cystoscopic Imaging for Bladder Cancer Detection Based on Stepwise Organic Transfer Learning with a Pretrained Convolutional Neural Network
Cystoscopic Imaging for Bladder Cancer Detection Based on Stepwise Organic Transfer Learning with a Pretrained Convolutional Neural Network
复制标题
DOI：
10.1089/end.2020.0919
复制
发表时间：
2020-12-04
影响因子：
2.7
通讯作者：
Nishiyama, Hiroyuki
中科院分区：
文献类型：
--
作者：
Ikeda, Atsushi;Nosato, Hirokazu;Nishiyama, Hiroyuki
背景：非肌层浸润性膀胱癌的诊断、治疗和监测均采用膀胱镜检查。人工智能（AI）越来越多地用于增强肿瘤检测，但其性能受到形成大型训练数据集所需的膀胱镜图像有限的阻碍。本研究旨在确定逐步迁移学习与一般图像，其次是胃镜图像可以提高膀胱肿瘤检测膀胱镜imaging.Materials和方法的准确性：我们训练了卷积神经网络与120万一般图像，其次是8728胃镜图像。在迁移学习过程的最后一步，该模型还使用在筑波大学医院收集的2102个正常膀胱组织和膀胱肿瘤的膀胱镜图像进行训练。采用受试者工作特征曲线评价诊断准确性。膀胱镜图像与或不逐步有机迁移学习训练的模型的诊断性能进行了比较，与医学生和泌尿科医生与不同层次的experience.Results：逐步有机迁移学习开发的模型有95.4%的灵敏度和97.6%的特异性。该性能优于其他模型，并与泌尿科专家的性能相当。值得注意的是，当肿瘤占据图像的>10%时，它显示出上级诊断准确性。结论：我们的研究结果证明了逐步有机迁移学习在有限数据集的应用中的价值，并进一步证实了AI在医学诊断中的价值。在这里，我们应用深度学习开发了一种检测膀胱肿瘤的工具，其准确性与泌尿科医生相当。为了解决很少有膀胱肿瘤图像可用于训练模型的局限性，我们证明了使用普通图像和胃镜图像进行预训练会产生上级结果。
Background: Nonmuscle-invasive bladder cancer is diagnosed, treated, and monitored using cystoscopy. Artificial intelligence (AI) is increasingly used to augment tumor detection, but its performance is hindered by the limited availability of cystoscopic images required to form a large training data set. This study aimed to determine whether stepwise transfer learning with general images followed by gastroscopic images can improve the accuracy of bladder tumor detection on cystoscopic imaging.Materials and Methods: We trained a convolutional neural network with 1.2 million general images, followed by 8728 gastroscopic images. In the final step of the transfer learning process, the model was additionally trained with 2102 cystoscopic images of normal bladder tissue and bladder tumors collected at the University of Tsukuba Hospital. The diagnostic accuracy was evaluated using a receiver operating characteristic curve. The diagnostic performance of the models trained with cystoscopic images with or without stepwise organic transfer learning was compared with that of medical students and urologists with varying levels of experience.Results: The model developed by stepwise organic transfer learning had 95.4% sensitivity and 97.6% specificity. This performance was better than that of the other models and comparable with that of expert urologists. Notably, it showed superior diagnostic accuracy when tumors occupied >10% of the image.Conclusions: Our findings demonstrate the value of stepwise organic transfer learning in applications with limited data sets for training and further confirm the value of AI in medical diagnostics. Here, we applied deep learning to develop a tool to detect bladder tumors with an accuracy comparable with that of a urologist. To address the limitation that few bladder tumor images are available to train the model, we demonstrate that pretraining with general and gastroscopic images yields superior results.