The SANDMED project aims at using the information acquired by whole body scanners, coupled, if possible, with other non-invasive acquisitions, to obtain relevant diagnostic parameters related to obesity and metabolic abnormalities (e.g., hyperinsulinaemia, hyperuricaemia, glucose intolerance, hypertension).
The basic idea of the project is to apply and customize recently developed shape analysis methods (e.g. PCA/LDA, spectral analysis, salient point detection and characterisation, skeleton-based segmentation and modelling) on a large set of acquired body scans to derive several compact parametric representations of the global and local variability of human body anatomy, testing the correlation between the parameters obtained and other diagnostic indicators.
In this way it will be possible to derive indices that are more effective than those commonly used now (e.g., BMI, W/H ratio), showing only a limited ability to predict medium- to long-term consequences of excess fat.
The project aims also at improving the acquisition protocols related to 3D whole body scanning and develop an interactive tool where not only high resolution surface models can be displayed and interactively analyzed, but they can also be enriched with computed indices and additional volumetric information coming from other modalities (e.g. US, DXA, MRI).