THREE DIMENSIONAL MAPPING FOR MELANOMA DIAGNOSIS
This medical device for melanoma screening can characterize and evaluate the potential risk of nevi based on data gathered through Structure-from-Motion (SfM) digital photogrammetry and infrared thermography. Thanks to Artificial intelligence the device associates a risk factor index to each detected skin lesion, thanks to the deep learning based on quantitative and qualitative information, three dimensional morphology, chromatic and thermographic data as well as the variation of such measurements over time.
The prototype, capable of acquiring data from the skin of a patient’s back, is currently in use for experimental clinical validation at the Istituto Oncologico Veneto Cancer Institute. The data acquired includes: photographic data; photogrammetric modelling, i.e. the 3D geometry of nevi with 0.2mm resolution; chromatic and thermal data; correlation of the thermographic data with the level of vascularisation; comparison with screenings from a previous examinations to identify variations. At the output the system provides a 2D model of the skin, with an indication of the melanoma risk index for each lesion identified, together with the data set of the suspicious melanomas. The risk index associated with each mole is determined by convolutional neural networks (CNNs) that classify the information acquired during screening based on information processed during training and on the comparison with previous screenings. The combination of these outputs allows a clear and objective assessment of the evolution of many nevi over time, as well as an evaluation of the risk factor of each mole thanks to the power of artificial intelligence.
- Nevi mapping for melanoma screening;
- Medical diagnostic device for clinics and hospitals;
- Dermatology analysis.
- Automated and fast Three dimensional nevi mapping;
- Includes thermal response / nevi vascularization data;
- Yields objective data necessary to monitor the evolution of nevi over time;
- Identifies nevi at risk of melanoma; Reduces human error;
- Data output in DICOM format;
- Continuous learning through AI;
- Saves money and effort on surgery and biopsies.