Classification of the Clinical Images for Benign and Malignant Cutaneous Tumors Using a Deep Learning Algorithm
Classification of the Clinical Images for Benign and Malignant Cutaneous Tumors Using a Deep Learning Algorithm
复制标题
DOI：
10.1016/j.jid.2018.01.028
复制
发表时间：
2018-07-01
影响因子：
6.5
通讯作者：
Chang, Sung Eun
中科院分区：
文献类型：
--
作者：
Han, Seung Seog;Kim, Myoung Shin;Chang, Sung Eun
我们测试了使用深度学习算法对12种皮肤疾病的临床图像进行分类-基底细胞癌，鳞状细胞癌，上皮内癌，光化性角化病，脂溢性角化病，恶性黑色素瘤，黑色素细胞痣，雀斑，化脓性肉芽肿，血管瘤，皮肤纤维瘤和疣。卷积神经网络（Microsoft ResNet-152模型; Microsoft Research Asia，中国北京）使用来自Asan数据集，MED-NODE数据集和图谱网站图像（总共19，398张图像）的训练部分的图像进行微调。使用Asan、Hallym和Edinburgh数据集的测试部分对训练模型进行验证。使用Asan数据集，诊断基底细胞癌、鳞状细胞癌、上皮内癌和黑色素瘤的曲线下面积分别为0.96 +/- 0.01、0.83 +/- 0.01、0.82 +/- 0.02和0.96 +/- 0.00。对于爱丁堡数据集，相应疾病的曲线下面积分别为0.90 +/- 0.01、0.91 +/- 0.01、0.83 +/- 0.01和0.88 +/- 0.01。使用Hallym数据集，基底细胞癌诊断的灵敏度为87.1% ± 6.0%。480 Asan和Edinburgh图像的测试算法性能与16名皮肤科医生的性能相当。为了提高卷积神经网络的性能，应该收集更多年龄和种族范围更广的图像。
We tested the use of a deep learning algorithm to classify the clinical images of 12 skin diseases-basal cell carcinoma, squamous cell carcinoma, intraepithelial carcinoma, actinic keratosis, seborrheic keratosis, malignant melanoma, melanocytic nevus, lentigo, pyogenic granuloma, hemangioma, dermatofibroma, and wart. The convolutional neural network (Microsoft ResNet-152 model; Microsoft Research Asia, Beijing, China) was fine-tuned with images from the training portion of the Asan dataset, MED-NODE dataset, and atlas site images (19,398 images in total). The trained model was validated with the testing portion of the Asan, Hallym and Edinburgh datasets. With the Asan dataset, the area under the curve for the diagnosis of basal cell carcinoma, squamous cell carcinoma, intraepithelial carcinoma, and melanoma was 0.96 +/- 0.01, 0.83 +/- 0.01, 0.82 +/- 0.02, and 0.96 +/- 0.00, respectively. With the Edinburgh dataset, the area under the curve for the corresponding diseases was 0.90 +/- 0.01, 0.91 +/- 0.01, 0.83 +/- 0.01, and 0.88 +/- 0.01, respectively. With the Hallym dataset, the sensitivity for basal cell carcinoma diagnosis was 87.1% +/- 6.0%. The tested algorithm performance with 480 Asan and Edinburgh images was comparable to that of 16 dermatologists. To improve the performance of convolutional neural network, additional images with a broader range of ages and ethnicities should be collected.