To date, majority of AI tools for skin cancer monitoring only assess single lesion images, in isolation of any patient clinical background.
The iToboS project aims to develop an innovative platform for skin cancer screening, harnessing AI-models to assess 3D total-body photography, along with patient clinical background to provide personalised monitoring approach. To achieve this, comprehensive training datasets consisting of patient images, genetic risk information and clinical history were required.
The clinical study protocol describes the methods for developing these training datasets to be used for machine learning. The protocol details methods for the acquisition of clinical data (including images, genetic risk, and medical history), and the annotation steps required for algorithm development. Recently, this protocol was published (peer-reviewed) with Frontiers in Medicine, in the spirit of increasing transparency, supporting reproducibility, and reducing publication bias.