Development of Subject-Specific Geometric Spine Model through Use of Automated Active Contour Segmentation and Kinematic Constraint-Limited Registration

Author： Strickland Catherine

Publisher： Springer Publishing Company

ISSN： 0897-1889

Source： Journal of Digital Imaging, Vol.24, Iss.5, 2011-10, pp. : 926-942

Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.

Previous Menu Next

Abstract

This paper describes the development of a patient-specific spine model through use of active contour segmentation and registration of intraoperative imaging of porcine vertebra augmented with kinematic constraints. The geometric active contours are fully automated and lead to a discrete representation of the image segmentation results. After determining errors within the segmentations, application of reliability theory allows the selection of active contour parameters to obtain best-fit segmentations from a stack of 2D images. The segmented images are then used in conjunction with C-arm fluoroscope images to simulate the result of intraoperative patient-specific model registration including patient and/or structure motion between preoperative and intraoperative scans. The results are validated through comparison of the error within the patient-specific model generated through use of the C-arm images with a model acquired directly from MRI images of the spine after motion. The results are applicable to the development of a wide variety of patient-specific geometric and biomechanical models.