Conformai mapping of the human scapula to generate dense landmark features

This study presents preliminary work in the detection of shape features through conformal mapping of the surface of the human scapula. The approach employs Ricci-flow based uniformization of the surface topology towards its canonical domain a sphere. The resulting evolution of the surface generates a distribution of conformal factors over the surface. The local maxima and minima of this distributed parameter are used as candidates for representations of local shape features. The procedure was tested on 5 scapulae and the detected features were compared to manual annotations 16 on each scapula. 3 out of 16 landmarks were closely approximated by the detected features with an average distance less than 2.1 mm. Visual inspection reveals other detected features that show apparent consistency in their anatomical location on the surface of the scapula.