Image processing and co-registration of CT and MRI Images: 3-D musculoskeletal geometry reconstruction for equine forelimb computer model generation.

Introduction: The methodology is described for 3-D musculoskeletal geometry reconstruction, required for computer model generation (SIMM®) of an equine forelimb, using computed tomography (CT), magnetic resonance imaging (MRI), and image analysis software (Analyze®). Methods: An isolated forelimb of a Thoroughbred racehorse was stabilized with a fiberglass cast in a Material Testing System to obtain a functional standing position. Reference markers detectable on MRI and CT were placed along the length of the dorsal and palmar surfaces of the cast leg along the mid-sagittal line. To enhance subsequent multimodal (MRI and CT) image registration two cylindrical markers were incorporated with the cast limb, secured to a PVC frame. CT and T1-weighted MRI data sets of the cast equine forelimb were acquired and downloaded to a UNIX workstation (SPARC Sun Station). An image analysis program (Analyze®) was used to co-register the two data sets and generate geometric coordinates of the markers, bones, muscles and ligaments. Manual segmentation of the muscles was obtained with a digitizing tablet (Drawing Slate II tablet, Calcomp Digitizer Division, 1994) and anatomic transverse sections of another equine forelimb were used to verify the anatomical identity and contours of the musculoskeletal structures during image processing and analysis. Results: Spline points of bone contours for each transverse slice of the CT/MRI registered file were obtained in Analyze® and used as the vertices describing the surface of the bone. A program was developed to connect the vertices for each bone to form the surface polygons required by the program (SIMM®) for the visualization of each bone's geometry. The 3-D volume for the nine muscle-tendon actuators was obtained after manual segmentation of the contour of each muscle and tendon from the transverse slices of the CT/MRI co-registered file. The origin and insertion points together with the centroid line of each muscle-tendon unit were calculated from the corresponding segmented muscle using Analyze® and similarly, for ligaments. Discussion: Co-registration of the CT/MRI images was essential in obtaining both data sets in the same generalized coordinate system (based on SIMM® requirements) and to better identify soft and hard tissue contours. Both semi-automatic and manual segmentation techniques were used to achieve muscle and ligament segmentation and subsequent 3D- reconstruction. Complete automatic soft tissue segmentation (i.e. muscles) is not yet available and despite advanced medical imaging and software systems ligament and muscle-tendon boundaries remain hard to define by simple thresholding.