An Intelligent Diagnostic Model for Melasma Based on Deep Learning and Multimode Image Input.
An Intelligent Diagnostic Model for Melasma Based on Deep Learning and Multimode Image Input.
复制标题
DOI：
10.1007/s13555-022-00874-z
复制
发表时间：
2023-03
影响因子：
3.4
通讯作者：
Chen, Jin
中科院分区：
文献类型：
--
作者：
Liu, Lin;Liang, Chen;Xue, Yuzhou;Chen, Tingqiao;Chen, Yangmei;Lan, Yufan;Wen, Jiamei;Shao, Xinyi;Chen, Jin
黄褐斑的诊断往往是基于医生的肉眼判断。然而，这对于缺乏经验的医生和非专业人士来说是一个挑战，不正确的治疗可能会产生严重的后果。因此，发展一种准确的黄褐斑诊断方法是非常重要的。本研究的目的是开发和验证基于深度学习的黄褐斑图像智能诊断系统。在VISIA系统中总共收集了8010张图像，包括4005张有黄褐斑患者的图像和4005张无黄褐斑患者的图像，用于训练和测试。受四种高性能结构的启发（即，DenseNet、ResNet、Swin Transformer和MobileNet），评估了深度学习模型在黄褐斑和非黄褐斑二进制分类器中的性能。此外，考虑到VISIA中每个镜头有五种图像模式，我们通过多通道图像输入以不同的组合融合这些模式，以探索多模式图像是否可以提高网络性能。基于DenseNet 121的网络在黄褐斑分类器的测试集上实现了最佳性能，准确率为93.68%，曲线下面积（AUC）为97.86%。加权类别激活图的结果表明，它是可解释的。在进一步的实验中，对于VISIA系统的五种模式，我们发现性能最好的模式是“BROWN SPOTS”。此外，“NORMAL”、“BROWN SPOTS”和“UV SPOTS”模式的组合显著提高了网络性能，实现了97.4%的最高准确度和99.28%的AUC。总之，深度学习对于诊断黄褐斑是可行的。该网络不仅对黄褐斑的临床图像具有良好的性能，而且可以在VISIA中使用多种模式的图像来获得高精度。在线版本包含补充材料，可通过10.1007/s13555-022-00874-z获得。
The diagnosis of melasma is often based on the naked-eye judgment of physicians. However, this is a challenge for inexperienced physicians and non-professionals, and incorrect treatment might have serious consequences. Therefore, it is important to develop an accurate method for melasma diagnosis. The objective of this study is to develop and validate an intelligent diagnostic system based on deep learning for melasma images. A total of 8010 images in the VISIA system, comprising 4005 images of patients with melasma and 4005 images of patients without melasma, were collected for training and testing. Inspired by four high-performance structures (i.e., DenseNet, ResNet, Swin Transformer, and MobileNet), the performances of deep learning models in melasma and non-melasma binary classifiers were evaluated. Furthermore, considering that there were five modes of images for each shot in VISIA, we fused these modes via multichannel image input in different combinations to explore whether multimode images could improve network performance. The proposed network based on DenseNet121 achieved the best performance with an accuracy of 93.68% and an area under the curve (AUC) of 97.86% on the test set for the melasma classifier. The results of the Gradient-weighted Class Activation Mapping showed that it was interpretable. In further experiments, for the five modes of the VISIA system, we found the best performing mode to be “BROWN SPOTS.” Additionally, the combination of “NORMAL,” “BROWN SPOTS,” and “UV SPOTS” modes significantly improved the network performance, achieving the highest accuracy of 97.4% and AUC of 99.28%. In summary, deep learning is feasible for diagnosing melasma. The proposed network not only has excellent performance with clinical images of melasma, but can also acquire high accuracy by using multiple modes of images in VISIA. The online version contains supplementary material available at 10.1007/s13555-022-00874-z.
登录
查看更多内容
DOI：
10.1109/tnnls.2022.3204090
发表时间：
2022-09-19
影响因子：
10.4
作者：
Du, Hao;Wang, Jiazheng;Meijering, Erik
通讯作者：
Meijering, Erik
影响因子：
7.7
作者：
Liang, Yunfeng;Sun, Lei;Lin, Zhiping
通讯作者：
Lin, Zhiping
DOI：
10.1364/josaa.415395
发表时间：
2021-06-01
影响因子：
1.9
作者：
Singh, Hardit;Saini, Simarjeet S.;Lakshminarayanan, Vasudevan
通讯作者：
Lakshminarayanan, Vasudevan
影响因子：
2.2
作者：
Lim, Ziying Vanessa;Akram, Farhan;Lee, Hwee Kuan
通讯作者：
Lee, Hwee Kuan
影响因子：
--
作者：
通讯作者：
--