CT-BASED VISUALIZATION AND SEGMENTATION OF SOFT TISSUE FEATURES FOR USE IN UPPER LIMB MODELING

Abstract

Biomechanical models have been successfully applied to screen potential risk factors for injuries and to plan and evaluate the effects of orthopedic surgical procedures.[1] These models have made apparent the feasibility and necessity for the generation of subject specific models that are aimed at custom clinical applications. In order to develop such models a methods needs to be developed that allows accurate geometrical visualization and reconstruction of position and characteristics of bone and soft tissues, including neurovascular structures.[2] In this study, we present our approach to obtain both bony as soft tissue features necessary for upper limb modeling from computer tomography alone. As a case study the techniques were applied in a non-anatomic shoulder reconstruction.

In order to determine the muscles of the shoulder girdle, ultrathin flexible metallic markers were sutured from origin to insertion according to the fiber directions in all muscles involved in shoulder movement on a total of ten different cadaver shoulders. The plexus brachialis and upper limb nerves were dissected and injected with a iodium contrast containing mixture. A Ct multi-slice image reconstruction was performed from occiput to the hip joint. The software package Mimics® (Materialise NV, Heverlee, Belgium) was used to segment and reconstruct the different anatomical models that included bone, muscle features, nerves and vascular structures. A clustering method algorithm, was used to filter interruptions of the different masks, scattering rustle and small irregularities due to the different contrasting markers used. Vascular tissue could be reconstructed and segmented as air filled structures. We were able to accurately reconstruct nerve tissue in an highly complex configuration such as the plexus brachialis.

Analysis of the representations showed that the different morphologic parameters were within the normal anatomical ranges and that our method is suitable to create complete anatomical models based on Ct-imaging alone.