Department of Biomedical Engineering, University of Basel, Basel, Switzerland.
Lucerne School of Computer Science and Information Technology, Lucerne University of Applied Sciences and Arts, Lucerne, Switzerland.
J Eur Acad Dermatol Venereol. 2022 Dec;36(12):2525-2532. doi: 10.1111/jdv.18476. Epub 2022 Sep 1.
The exact location of skin lesions is key in clinical dermatology. On one hand, it supports differential diagnosis (DD) since most skin conditions have specific predilection sites. On the other hand, location matters for dermatosurgical interventions. In practice, lesion evaluation is not well standardized and anatomical descriptions vary or lack altogether. Automated determination of anatomical location could benefit both situations.
Establish an automated method to determine anatomical regions in clinical patient pictures and evaluate the gain in DD performance of a deep learning model (DLM) when trained with lesion locations and images.
Retrospective study based on three datasets: macro-anatomy for the main body regions with 6000 patient pictures partially labelled by a student, micro-anatomy for the ear region with 182 pictures labelled by a student and DD with 3347 pictures of 16 diseases determined by dermatologists in clinical settings. For each dataset, a DLM was trained and evaluated on an independent test set. The primary outcome measures were the precision and sensitivity with 95% CI. For DD, we compared the performance of a DLM trained with lesion pictures only with a DLM trained with both pictures and locations.
The average precision and sensitivity were 85% (CI 84-86), 84% (CI 83-85) for macro-anatomy, 81% (CI 80-83), 80% (CI 77-83) for micro-anatomy and 82% (CI 78-85), 81% (CI 77-84) for DD. We observed an improvement in DD performance of 6% (McNemar test P-value 0.0009) for both average precision and sensitivity when training with both lesion pictures and locations.
Including location can be beneficial for DD DLM performance. The proposed method can generate body region maps from patient pictures and even reach surgery relevant anatomical precision, e.g. the ear region. Our method enables automated search of large clinical databases and make targeted anatomical image retrieval possible.
皮肤损伤的确切位置在临床皮肤科中至关重要。一方面，它支持鉴别诊断（DD），因为大多数皮肤疾病都有特定的好发部位。另一方面，位置对于皮肤科手术干预也很重要。在实践中，病变评估没有得到很好的标准化，解剖描述要么存在差异，要么完全缺乏。自动确定解剖位置可以同时受益于这两种情况。
建立一种自动确定临床患者图像中解剖区域的方法，并评估将病变位置和图像纳入深度学习模型（DLM）训练中对 DD 性能的增益。
基于三个数据集进行回顾性研究：宏观解剖学用于身体主要区域，有 6000 张患者图片由一名学生进行部分标记；微观解剖学用于耳部区域，有 182 张图片由一名学生标记；DD 则有 3347 张由皮肤科医生在临床环境下确定的 16 种疾病的图片。为每个数据集训练和评估了一个 DLM。主要的评估指标是 95%置信区间（CI）内的精度和敏感度。对于 DD，我们比较了仅使用病变图片训练的 DLM 和同时使用图片和位置训练的 DLM 的性能。
宏观解剖学的平均精度和敏感度分别为 85%（CI 84-86）和 84%（CI 83-85）；微观解剖学的平均精度和敏感度分别为 81%（CI 80-83）和 80%（CI 77-83）；DD 的平均精度和敏感度分别为 82%（CI 78-85）和 81%（CI 77-84）。我们观察到，当同时使用病变图片和位置进行训练时，DD 的平均精度和敏感度分别提高了 6%（McNemar 检验 P 值<0.0009）。
包括位置信息可能对 DD DLM 的性能有益。所提出的方法可以从患者图片中生成身体区域图，甚至可以达到手术相关的解剖精度，例如耳部区域。我们的方法能够自动搜索大型临床数据库，并实现有针对性的解剖图像检索。
