Melanoma is one of the most aggressive forms of skin cancer, responsible for 60% of lethal skin neoplasms. Early detection is crucial for improving patient outcomes, particularly as the incidence of melanoma continues to rise, posing an increasing public health challenge due to the aging population and extended life expectancy.
Over the past few decades, dermoscopy has revolutionized the diagnosis of skin lesions by enabling in vivo observation of morphological structures not visible to the naked eye, many of which correlate with histopathological findings.
As such, dermoscopy has been incorporated into the total body skin examination as the primary screening method for melanoma, which involves examining each pigmented skin lesion for typical signs of melanoma. This has led to the adoption and standardization of digital follow-up techniques such as the two-step method, which combines total-body photograph (TBP) and sequential digital dermoscopy imaging (SDDI). By allowing clinicians to compare current images with previous records, these tools enable the early detection of new or evolving lesions, as well as the monitoring of macroscopic and dermoscopic changes in pre-existing ones.
Although this method is highly effective, it can be time-consuming and is therefore selectively aimed at groups with high-risk factors, such as patients presenting a large number of nevi or atypical mole syndrome.
To optimize both the efficacy and efficiency of total body examination, the integration of artificial intelligence (AI) in both TBP and dermoscopy is emerging as a promising innovation. The Intelligent Total Body Scanner for Early Detection of Melanoma (iToBoS) project aims to lead the way forward in this direction, developing an AI-based diagnostic platform that integrates various data sources including medical records, genomics data, and imaging. It does not rely solely on image analysis but also contributes to registering and eventually defining several different patient phenotypes based on various types of risk factors. Hence, this platform aims to provide a highly personalized risk assessment for each patient and for each lesion following a more global and phenotype-based approach. It will ultimately offer healthcare practitioners enhanced diagnostic tools while addressing the "black box" issue of AI models through improved transparency and interpretability.
This article reviews the current standards in digital follow-up and the application of AI within the iToBoS project. It covers the indications for digital monitoring in high-risk populations, details the two-step follow-up method, provides a technical overview of imaging acquisition, and discusses the interpretation of processed images. Finally, it reviews the implications of several hardware and software technologies, including AI algorithms in the iToBoS project.
Continue learning through the iToBoS whitepaper "Standards in whole body photography, digital dermoscopy and artificial intelligence applied to the early diagnosis of melanoma".