Rigid US-MRI registration through segmentation of equivalent anatomic structures a feasibility study using 3D transcranial ultrasound of the midbrain

Abstract

Multi-modal registration between 3D ultrasound (US) and magnetic resonance imaging (MRI) is motivated by aims such as image fusion for improved diagnostics or intra-operative evaluation of brain shift. In this work, we present a rigid region-based registration approach between MRI and 3D-US based on the segmentation of equivalent anatomic structures in both modalities. Our feasibility study is performed using segmentations of the midbrain in both MRI and 3D transcranial ultrasound. Segmentation of MRI is based on deformable atlas registration while for 3D US segmentation, we recently proposed an accurate and robust method based on statistical shape modeling and a discrete and localized active surface segmentation framework. The multimodal registration is performed through intensity-based rigid registration of signed distance transforms of both segmentations. Qualitative results and a demonstration of the basic feasibility of the region-based registration are demonstrated on a pair of MRI and challenging 3D transcranial US data volumes from the same subject. © Springer-Verlag Berlin Heidelberg 2012.