背景:人工智能(AI)正在重塑医疗保健,使用机器和深度学习来加强疾病管理。皮肤科已经改善了诊断,特别是在皮肤癌检测中,通过整合AI。然而,AI在自身免疫性大疱性皮肤病的免疫荧光成像自动化方面的潜力仍未开发.虽然直接免疫荧光(DIF)支持诊断,它的人工解释会妨碍效率。使用深度学习对DIF模式进行自动分类,包括细胞间模式(ICP)和线性模式(LP),有望改善自身免疫性大疱性皮肤病的诊断。
目的:本研究的目的是开发用于自身免疫性大疱性皮肤病DIF模式自动分类的AI算法,如ICP和LP。这旨在提高诊断准确性,简化疾病管理,并通过深度学习驱动的免疫荧光解释改善患者的预后。
方法:我们收集了2022年1月至2024年1月期间疑似AIBD患者皮肤活检的免疫荧光图像。通过5mm穿刺活检获得皮肤组织,为直接免疫荧光准备。经验丰富的皮肤科医生将图像分为三类:ICP,LP,和消极。为了评估我们的深度学习方法,我们将图像分为训练集(436)和测试集(93)。我们采用了预训练深度神经网络的迁移学习,并进行了5倍交叉验证来评估模型性能。我们的数据集的类不平衡是使用加权损失和数据增加策略解决的。使用Pytorch对模型进行了50个时期的训练,为CNN和SwinTransformer实现224x224的图像大小。
结果:我们的研究比较了六个CNN和用于AIBDs图像分类的Swin变换器,Swin变压器实现了98.5%的最高平均验证精度。在单独的测试集上,最佳模型达到了94.6%的准确率,在AIBDs类别中表现出95.3%的敏感性和97.5%的特异性。使用Grad-CAM进行的可视化突出了模型对特征模式的依赖,以实现准确的分类。
结论:该研究强调了CNN在识别DIF特征方面的准确性。这种方法有助于自动分析和报告,提供再现性,速度,数据处理,和成本效益。在皮肤免疫荧光中集成深度学习有望在皮肤科的这一分支中进行精确的诊断和简化的报告。
BACKGROUND: Artificial intelligence (AI) is reshaping healthcare, using machine and deep learning (DL) to enhance disease management. Dermatology has seen improved diagnostics, particularly in skin cancer detection, through the integration of AI. However, the potential of AI in automating immunofluorescence imaging for autoimmune bullous skin diseases (AIBDs) remains untapped. While direct immunofluorescence (DIF) supports diagnosis, its manual interpretation can hinder efficiency. The use of DL to classify DIF patterns automatically, including the intercellular (ICP) and linear pattern (LP), holds promise for improving the diagnosis of AIBDs.
OBJECTIVE: To develop AI algorithms for automated classification of AIBD DIF patterns, such as ICP and LP, in order to enhance diagnostic accuracy, streamline disease management and improve patient outcomes through DL-driven immunofluorescence interpretation.
METHODS: We collected immunofluorescence images from skin biopsies of patients suspected of having an AIBD between January 2022 and January 2024. Skin tissue was obtained via a 5-mm punch biopsy, prepared for DIF. Experienced dermatologists classified the images as ICP, LP or negative. To evaluate our DL approach, we divided the images into training (n = 436) and test sets (n = 93). We employed transfer learning with pretrained deep neural networks and conducted fivefold cross-validation to assess model performance. Our dataset\'s class imbalance was addressed using weighted loss and data augmentation strategies. The models were trained for 50 epochs using Pytorch, achieving an image size of 224 × 224 pixels for both convolutional neural networks (CNNs) and the Swin Transformer.
RESULTS: Our study compared six CNNs and the Swin Transformer for AIBD image classification, with the Swin Transformer achieving the highest average validation accuracy (98.5%). On a separate test set, the best model attained an accuracy of 94.6%, demonstrating 95.3% sensitivity and 97.5% specificity across AIBD classes. Visualization with Grad-CAM (class activation mapping) highlighted the model\'s reliance on characteristic patterns for accurate classification.
CONCLUSIONS: The study highlighted the accuracy of CNNs in identifying DIF features. This approach aids automated analysis and reporting, offering reproducibility, speed, data handling and cost-efficiency. Integrating DL into skin immunofluorescence promises precise diagnostics and streamlined reporting in this branch of dermatology.
Artificial intelligence (AI) is transforming healthcare through machine and deep learning (computer systems that can learn and adapt, and make complex decisions, without receiving explicit instructions), improving disease management in dermatology, particularly in detecting skin cancer. However, AI’s potential in automating immunofluorescence imaging in autoimmune bullous (blistering) skin diseases (AIBDs) remains largely untapped. Manual interpretation of direct immunofluorescence (DIF – a type of microscopy) can reduce efficiency. However, using deep learning to automatically classify DIF patterns (for example, the ‘intercellular pattern’ (ICP) and the ‘linear pattern’ (LP)) holds promise in helping with the diagnosis of AIBDs. This study aimed to develop AI algorithms for the automated classification of AIBD DIF patterns, such as ICP and LP, to improve diagnostic accuracy and streamline disease management. Immunofluorescence images were collected from skin biopsies of patients with a suspected AIBD between January 2022 and January 2024. Dermatologists classified the images into three categories: ICP, LP and negative. The dataset was divided into training (436 images) and test sets (93 images). A transfer learning framework (where what has been learned previously in one setting is used to improve performance in another) was used to make up for the limited amount of training data, to explore different models for the AIBD classification task. Our results revealed that a model called the ‘Swin Transformer’ achieved an average accuracy of 99% in diagnosing different AIBDs. The best model attained 95% accuracy on the test set and was reliable in identifying and ruling out different AIBDs. Visualization with Grad-CAM (a technique used in deep learning) highlighted the model’s use of characteristic patterns to classify the diseases accurately. Overall, integrating deep learning in skin immunofluorescence promises to improve diagnostics and streamline reporting in dermatology, which could improve consistency, speed and cost-efficiency.