Yu Shijuan, Chen Zhilin, He Jingyi, Wang Hua
Department of Dermatology, Children's Hospital of Chongqing Medical University, National Clinical Research Center for Child Health and Disorders, Ministry of Education Key Laboratory of Child Development and Disorders, China International Science and Technology Cooperation base of Child development and Critical Disorders, Chongqing Key Laboratory of Child Infection and Immunity, Chongqing 400014, China; The Seventh People's Hospital of Chongqing, Chongqing 400054, China.
Software Engineering, International College of Chongqing University of Posts and Telecommunications, No. 2 Chongwen Road, Nan'an District, Chongqing 400065, China.
Photodiagnosis Photodyn Ther. 2025 Aug;54:104727. doi: 10.1016/j.pdpdt.2025.104727. Epub 2025 Jul 16.
To explore the performance of a deep learning (DL) model based on dermoscopy images in diagnosing childhood vitiligo.
A total of 474 pediatric patients (223 with vitiligo and 251 without vitiligo) were enrolled. Three types of imaging data were collected: dermoscopic images, Wood's lamp images, and standard clinical photographs. Two diagnostic evaluation approaches were established. Clinician-based assessment: Eight dermatologists performed a double-blind evaluation using dermoscopic images. DL-based assessment: ResNet152 and DenseNet121 models were trained on 3896 dermoscopic images (with an 8:2 split between the development set and validation set). The evaluation metrics included the AUC of ROC curve, sensitivity, specificity, F1-score, and accuracy. Additionally, the correlation between clinicians' diagnostic performance and their years of experience was analyzed.
ROC curve analysis revealed that using the training questionnaire as a control group, the diagnostic performance of dermatologists for vitiligo based solely on dermoscopy images yielded an AUC of 0.77 (95 % CI: 0.51-1.00), sensitivity of 0.88 (95 % CI: 0.53-0.99), and specificity of 0.75 (95 % CI: 0.41-0.96). The confusion matrix for the ResNet152 model indicated an accuracy of 83.08 %, a recall rate of 86.84 %, a precision of 81.08 %, a specificity of 79.22 %, an F1 score of 0.8386, and an AUC of 0.91. The confusion matrix for the DenseNet121 model indicated an accuracy of 81.41 % and a recall rate of 83.41 % (precision: 82.03 %, specificity: 79.12 %, F1 score: 0.8271, and AUC: 0.89).
Both DL models based on dermoscopy images exhibit high overall classification performance in the diagnosis of childhood vitiligo.
探讨基于皮肤镜图像的深度学习（DL）模型在儿童白癜风诊断中的性能。
共纳入474例儿科患者（223例白癜风患者和251例非白癜风患者）。收集了三种类型的影像数据：皮肤镜图像、伍德灯图像和标准临床照片。建立了两种诊断评估方法。基于临床医生的评估：八位皮肤科医生使用皮肤镜图像进行双盲评估。基于DL的评估：在3896张皮肤镜图像上训练ResNet152和DenseNet121模型（开发集和验证集按8:2划分）。评估指标包括ROC曲线的AUC、敏感性、特异性、F1分数和准确性。此外，分析了临床医生诊断性能与其经验年限之间的相关性。
ROC曲线分析显示，以训练问卷作为对照组，皮肤科医生仅基于皮肤镜图像诊断白癜风的性能，AUC为0.77（95%CI：0.51 - 1.00），敏感性为0.88（95%CI：0.53 - 0.99），特异性为0.75（95%CI：0.41 - 0.96）。ResNet152模型的混淆矩阵显示准确率为83.08%，召回率为86.84%，精确率为81.08%，特异性为79.22%，F1分数为0.8386，AUC为0.91。DenseNet121模型的混淆矩阵显示准确率为81.41%，召回率为83.41%（精确率：82.03%，特异性：79.12%，F1分数：0.8271，AUC：0.89）。
基于皮肤镜图像的两种DL模型在儿童白癜风诊断中均表现出较高的总体分类性能。
