SkinDiag

Skin cancers could be aggressive with increasing incidence, representing a major public health challenge. Although computer-aided diagnosis (CAD) systems based on dermoscopic images and deep learning have shown high performance, most existing approaches rely on the analysis of isolated lesions and neglect the fundamental dermatological concepts and global reasoning processes, and the aid of multimodal approach used by clinicians in real practice. This gap limits both their clinical relevance and interpretability.

This interdisciplinary PhD project aims to deepen the modeling of dermatological concepts for skin cancers diagnosis, basal cell carcinoma by the multi modal approach and melanoma diagnosis by analyzing the entirety of a patient’s skin lesions rather than single images. This objective is enabled by the VECTRA 3D total body mapping system recently installed at La Conception Dermatology Department, which provides high-resolution, standardized images of the full skin surface and all identified lesions. Skin cancers can be diagnosed based on a multimodal imaging bay including VECTRA 3D system, videodermosocpy and AI, Reflectance Confocal Microscopy (RCM) and LC-OCT.