MRI-based 3D-Modelling of Gleno-humeral joint deformities for functional surgical planning

Abstract

The evaluation of gleno-humeral joint congruency and dysplasia is mandatory before reconstructive surgery in children suffering from severe upper obstetric brachial plexus palsy. Magnetic resonance imaging (MRI) is used for the assessment of the joint incongruency and the state of the surrounding soft tissue. 3D morphological and functional information is needed for an enhanced preoperative surgical planning. For accurate functional modelling, however, the amount of patient-specific information should be maximized. In this paper we present an approach for obtaining patient-specific morphological models from shoulder MR-images acquired under clinical conditions. Referential image data was used to reconstruct a generic anatomical model that provides morphological description of gleno-humeral joint and muscles attachments as one prerequisite for enhanced functional modelling. A non-rigid registration method is used to adapt the generic model to the patient's image data by deformation algorithms. Few anatomical landmarks were manually defined in the region of interest by a medical expert and have been used for a global registration step using the thin plate spline (TPS) transformation. Image enhancement and segmentation algorithms have been developed to extract features on the bony structure's surface. The final fitting of the model was then achieved after computing the smooth non-rigid transformation between the model and features extracted from the patient images using the coherent point drift (CPD) method. Application of the proposed approach on clinical MR-images is presented. © 2009 Springer Berlin Heidelberg.