Bone microstructure reconstructions from few projections with level-set regularization

In this work, we study a new reconstruction method for binary tomography cross-sections of the bone microstructure. The binary cross-sections are obtained from a small number of projections and the binary tomography problem is formulated as a non linear inverse problem. Stable solution are obtained with level-set methods and a Bounded Variation-Sobolev regularization of the level set function. The method is applied to 256 × 256 images and with additive gaussian noise. Good results are are obtained from only 15-20 projections.