Artificial intelligence in three-dimensional total-body photography for skin cancer surveillance

Artificial intelligence is often used in studies that detect skin cancer, but most work uses images of selected lesions. These images are useful for classification. However, they do not fully match real screening situations. In lesion surveillance, clinicians need to inspect many lesions across the whole body and monitor whether lesions change over time. Three-dimensional total-body photography is able to capture a wider skin surface and can support lesion selection, lesion triage, risk assessment, and follow-up comparison. This approach gives artificial intelligence more opportunity to analyze the patient beyond one selected lesion. This Mini Review discusses recent evidence on artificial intelligence-assisted three-dimensional total-body photography for skin cancer surveillance. In current applications, this approach is also used in automated triage and lesion detection, multimodal risk prediction, phenotype extraction, and longitudinal tracking. The reviewed papers show early progress, but the evidence remains limited. Skin tone reporting, workflow integration, dataset transparency, false-positive and false-negative harm, cost, and equity remain key adoption issues. The objective is to focus on the shift from selected-lesion classification to whole-body imaging. Clinical value should not be judged only by lesion-level accuracy, but also by whether artificial intelligence-assisted three-dimensional total-body photography can improve patient-level surveillance across the whole skin surface and over time. Stronger prospective, longitudinal, workflow, cost, and equity evidence is still needed before routine clinical adoption in practice.