基于深度学习的肠腺瘤病变识别
|
摘要
目的建立人工智能辅助诊断肠腺瘤诊断系统。方法我们筛选出近187例覆盖各种肠腺瘤组织形态的病理切片,利用数字扫描仪将其数字化后,医生借助ASAP标注工具对数字病理切片进行标注。标注完成后,我们对标注数据进行处理并分割,得到超过150万张带有标注的训练数据,最后输入到卷积神经网络中进行模型的训练。结果基于训练完成的深度学习模型,我们在155例切片的测试集上进行测试,模型可以达到94.8%的准确率。结论腺瘤的诊断是人工智能在病理诊断中较为简单的一个模型,随着人工智能技术的发展,基于人工智能的病理辅助诊断技术必将极大地解放病理医生的体力,促进病理学的发展。
Objective To establish an artificial intelligence(AI)-assisted diagnosis system for intestinal adenoma. Methods We selected 187 pathological slides covering the morphology of different intestinal adenomas. After digitized with a scanner, the slides were labeled by pathologists with the ASAP annotation tool. After the annotation was completed, we preprocessed and segmented the slides to obtain more than 1.5 million labeled training data, and fed them into the convolutional neural network. Results We tested the performance of the trained deep learning model on the test set containing 155 slices, and our model achieved an accuracy of 94.8%. Conclusion The diagnosis of intestinal adenoma is a starting point of applying AI to pathological diagnosis. With the development of deep learning, AI-assisted pathological diagnosis will significantly reduce the burden of pathologists and promote the development of pathology.
Objective To establish an artificial intelligence(AI)-assisted diagnosis system for intestinal adenoma. Methods We selected 187 pathological slides covering the morphology of different intestinal adenomas. After digitized with a scanner, the slides were labeled by pathologists with the ASAP annotation tool. After the annotation was completed, we preprocessed and segmented the slides to obtain more than 1.5 million labeled training data, and fed them into the convolutional neural network. Results We tested the performance of the trained deep learning model on the test set containing 155 slices, and our model achieved an accuracy of 94.8%. Conclusion The diagnosis of intestinal adenoma is a starting point of applying AI to pathological diagnosis. With the development of deep learning, AI-assisted pathological diagnosis will significantly reduce the burden of pathologists and promote the development of pathology.
引文
[1] 包骥,步宏. 中国数字病理发展展望[J]. 实用医院临床杂志,2017, 14 (5):1-2.
[2] 邓杨,包骥. 数字病理中计算机辅助诊断研究展望[J]. 实用医院临床杂志, 2017,14(5):10-12.
[3] Krizhevsky A, Sutskever I. Imagenet classification with deep convolutional neural networks[J].Adv Neural Inf Process Syst, 2012, 25:1106-1114.
[4] Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]//International Conference on Machine Learning. JMLR, 2015:448-456.
[5] Russakovsky O, Deng J, Su H, et al. Image net large scale visual recognition challenge[J]. Int J Comput Vis, 2014, 115(3):211-252.
[6] Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2014:580-587.
[7] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2015:3431-3440.
[8] 陈梅,吕晓娟,张麟. 人工智能助力医疗的机遇与挑战[J]. 中国数字医学, 2018,13(1):16-18.
[9] Mart?n Abadi, Ashish Agarwal, Paul Barham, et al. TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems[OL]. www.researchgate.net/publication/301839500. 2016
[10] 宋麦玲,李欢. 一种基于HSV颜色空间的图像检索技术[J]. 电脑知识与技术, 2007, 1 (1):200.
[11] 刘步实,刘致锦,覃晓. 随机森林在图像分割中的应用研究[J]. 现代计算机, 2017,13:3-6.
[12] Sari CT, Gunduz-Demir C. Unsupervised feature extraction via deep learning for histopathological classification of colon tissue images[J]. IEEE Trans Med Imaging, 2018,暂未分卷.
[13] Xu S, Tai D, Wee A, et al. Automated scoring of liver fibrosis through combined features from different collagen groups[J]. Conf IEEE Eng Med Biol Soc, 2011,2011:4503-4506.
[14] Arvaniti E, Fricker KS, Moret M, et al. Automated gleason grading of prostate cancer tissue microarrays via deep learning[J]. Sci Rep, 2018, 8(1):12054.
[15] Steiner DF, Macdonald R, Liu Y, et al. Impact of deep learning assistance on the histopathologic review of lymph nodes for metastatic breast cancer[J]. Am Surg Pathol, 2018, 42(12):1636-1646.
[16] Han I, Kim JH, Park H, et al. Deep learning approach for survival prediction for patients with synovial sarcoma[J]. Tumour Biol, 2018, 40(9):1010428318799264.
[17] Rawat RR, Ruderman D, Macklin P, et al. Correlating nuclear morphometric patterns with estrogen receptor status in breast cancer pathologic specimens[J]. NPJ Breast Cancer, 2018,4:32.
[18] Couture HD, Williams LA, Geradts J, et al. Image analysis with deep learning to predict breast cancer grade, ER status, histologic subtype, and intrinsic subtype[J]. NPJ Breast Cancer, 2018,4:30.
[19] Coudray N, Ocampo PS, Sakellaropoulos T, et al.Classification and mutation prediction from non-small cell lung cancer histopathology images using deep learning[J]. Nat Med, 2018, 24(10):1559-1567.
[2] 邓杨,包骥. 数字病理中计算机辅助诊断研究展望[J]. 实用医院临床杂志, 2017,14(5):10-12.
[3] Krizhevsky A, Sutskever I. Imagenet classification with deep convolutional neural networks[J].Adv Neural Inf Process Syst, 2012, 25:1106-1114.
[4] Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]//International Conference on Machine Learning. JMLR, 2015:448-456.
[5] Russakovsky O, Deng J, Su H, et al. Image net large scale visual recognition challenge[J]. Int J Comput Vis, 2014, 115(3):211-252.
[6] Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2014:580-587.
[7] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2015:3431-3440.
[8] 陈梅,吕晓娟,张麟. 人工智能助力医疗的机遇与挑战[J]. 中国数字医学, 2018,13(1):16-18.
[9] Mart?n Abadi, Ashish Agarwal, Paul Barham, et al. TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems[OL]. www.researchgate.net/publication/301839500. 2016
[10] 宋麦玲,李欢. 一种基于HSV颜色空间的图像检索技术[J]. 电脑知识与技术, 2007, 1 (1):200.
[11] 刘步实,刘致锦,覃晓. 随机森林在图像分割中的应用研究[J]. 现代计算机, 2017,13:3-6.
[12] Sari CT, Gunduz-Demir C. Unsupervised feature extraction via deep learning for histopathological classification of colon tissue images[J]. IEEE Trans Med Imaging, 2018,暂未分卷.
[13] Xu S, Tai D, Wee A, et al. Automated scoring of liver fibrosis through combined features from different collagen groups[J]. Conf IEEE Eng Med Biol Soc, 2011,2011:4503-4506.
[14] Arvaniti E, Fricker KS, Moret M, et al. Automated gleason grading of prostate cancer tissue microarrays via deep learning[J]. Sci Rep, 2018, 8(1):12054.
[15] Steiner DF, Macdonald R, Liu Y, et al. Impact of deep learning assistance on the histopathologic review of lymph nodes for metastatic breast cancer[J]. Am Surg Pathol, 2018, 42(12):1636-1646.
[16] Han I, Kim JH, Park H, et al. Deep learning approach for survival prediction for patients with synovial sarcoma[J]. Tumour Biol, 2018, 40(9):1010428318799264.
[17] Rawat RR, Ruderman D, Macklin P, et al. Correlating nuclear morphometric patterns with estrogen receptor status in breast cancer pathologic specimens[J]. NPJ Breast Cancer, 2018,4:32.
[18] Couture HD, Williams LA, Geradts J, et al. Image analysis with deep learning to predict breast cancer grade, ER status, histologic subtype, and intrinsic subtype[J]. NPJ Breast Cancer, 2018,4:30.
[19] Coudray N, Ocampo PS, Sakellaropoulos T, et al.Classification and mutation prediction from non-small cell lung cancer histopathology images using deep learning[J]. Nat Med, 2018, 24(10):1559-1567.