Landmarking-based unsupervised clustering of human faces manifesting labio-schisis dysmorphisms

Ultrasound scans, Computed Axial Tomography, Magnetic Resonance Imaging are only few examples of medical imaging tools boosting physicians in diagnosing a wide range of pathologies. Anyway, no standard methodology has been defined yet to extensively exploit them and current diagnoses procedures are still carried out mainly relying on physician's experience. Although the human contribution is always fundamental, it is self-evident that an automatic procedure for image analysis would allow a more rapid and effective identification of dysmorphisms. Moving toward this purpose, in this work we address the problem of feature extraction devoted to the detection of specific diseases involving facial dysmorphisms. In particular, a bounded Depth Minimum Steiner Trees (D-MST) clustering algorithm is presented for discriminating groups of individuals relying on the manifestation/absence of the labio-schisis pathology, commonly called cleft lip. The analysis of three-dimensional facial surfaces via Differential Geometry is adopted to extract landmarks. The extracted geometrical information is furthermore elaborated to feed the unsupervised clustering algorithm and produce the classification. The clustering returns the probability of being affected by the pathology, allowing physicians to focus their attention on risky individuals for further analysis.