一种基于增强卷积神经网络的病理图像诊断算法 下载: 1243次
Algorithm for Pathological Image Diagnosis Based on Boosting Convolutional Neural Network
天津大学电气自动化与信息工程学院, 天津 300072
图 & 表
图 1. 增强卷积网络模型的训练过程
Fig. 1. Training process of boosting convolutional neural network model
下载图片 查看原文
图 2. 增强卷积网络模型的测试过程
Fig. 2. Test process of boosting convolutional neural network model
下载图片 查看原文
表 1肾、肺与脾的图像块级与图级诊断准确率
Table1. Patch-level and image-level diagnosis accuracies of kidney, lung and spleen%
Method | Kidney | Lung | Spleen | | |
---|
Patch | Image | | | Patch | Image | Patch | Image |
---|
Basic classifier | 82.80 | 92.73 | 91.31 | 98.18 | 90.98 | 98.18 | Proposed | 84.15 | 93.64 | 93.31 | 99.09 | 92.10 | 98.18 |
|
查看原文
表 2肾、肺与脾图像的诊断结果
Table2. Diagnosis results of kidney, lung and spleen images%
Class | Kidney | Lung | Spleen | Method |
---|
Healthy | Inflammatory | | Healthy | Inflammatory | Healthy | Inflammatory |
---|
| 88.19 | 11.81 | 84.55 | 15.45 | 91.47 | 8.53 | PoE | Healthy | 88.21 | 11.79 | 96.52 | 3.48 | 92.88 | 7.12 | DFDL | | 96.36 | 3.64 | 99.09 | 0.91 | 100.00 | 0.00 | Proposed | | 14.96 | 85.04 | 7.32 | 92.68 | 10.08 | 89.92 | PoE | Inflammatory | 9.92 | 90.08 | 2.57 | 97.43 | 7.89 | 92.11 | DFDL | | 7.28 | 92.72 | 0.00 | 100.00 | 1.82 | 98.18 | Proposed |
|
查看原文
表 3Camelyon16数据集中图像的块级诊断准确率
Table3. Patch-level diagnosis accuracies of images in Camelyon16 dataset%
Method | Accuracy |
---|
IMBC | 92.1 | IMSLN | 92.7 | Proposed | 93.3 |
|
查看原文
孟婷, 刘宇航, 张凯昱. 一种基于增强卷积神经网络的病理图像诊断算法[J]. 激光与光电子学进展, 2019, 56(8): 081001. Ting Meng, Yuhang Liu, Kaiyu Zhang. Algorithm for Pathological Image Diagnosis Based on Boosting Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2019, 56(8): 081001.