Skeletonization by blocks for large 3D datasets: application to brain microcirculation

Skeletons are compact representations that allow mathematical analysis of objects. A skeleton must be homotopic, thin and medial in relation to the object it represents. Numerous approaches already exist which focus on computational efficiency. However, when dealing with data too large to be loaded into the main memory of a personal computer, such approaches can no longer be used. We present in this article a skeletonization algorithm that processes the data locally (in sub-images) while preserving global properties (medial localization). Our privileged application is the study of the cerebral micro-vascularisation, and we show some results obtained on a mosaic of 3-D images acquired by confocal microscopy.