Patient-specific technology for in vivo assessment of 3-D spinal motion

One of the most common musculoskeletal problems affecting people is neck and low back pain. Traditional clinical diagnostic techniques such as fluoroscopic imaging or CT scans are limited due to their static and/or planar measurements which may not be able to capture all neurological pathologies. More advanced diagnostics have proven successful in assessing 3-D patient-specific spinal kinematics by combining a patient-specific 3-D spine model (CT or MRI) with bi-planar fluoroscopic imaging; however, custom, not clinically available advanced imaging equipment as well as an increase in radiation exposure is required to acquire a complete patient-specific spinal kinematic description. Hence, the purpose of this research was to develop a clinically viable bi-planar fluoroscopic imaging technique which acquires a complete patient-specific kinematic description of the spine with reduced radiation exposure. Development of the proposed technique required evaluating the accuracy of 3-D kinematic interpolation techniques in reconstructing spinal kinematic data in order to reduce radiation exposure from bi-planar fluoroscopic diagnostic techniques. Several interpolation and sampling algorithms were evaluated in reconstructing cadaveric lumbar (L2-S1) flexion-extension motion data; ultimately, a new interpolation algorithm was proposed. Similarly, the success of the interpolation algorithm was evaluated in reconstructing spine-specific kinematic parameters. Next, the interpolation algorithm was combined with a CT-based bi-planar fluoroscopic method. Accuracy of the proposed diagnostic technique was evaluated against previously validated work on an ex vivo optoelectronic 3-D kinematic assessment technique. Bi-planar fluoroscopic images were acquired during both flexion-extension and lateral bending motions of cadaveric cervical (C4-T1) and lumbar (L2-S1) spine. Registration of the bi-planar fluoroscopic images to the CT-based 3-D model was optimized using a gradient derived similarity function. Additionally, a stochastic approach, covariance matrix adaptive evolution strategy, was used as the optimizing function. The newly developed interpolation algorithm was used to reduce the sample size of the bi-planar fluoroscopic images which reduces radiation exposure. Experimental results illustrate the potential success of the technique, but ultimately improvements in registration and validation methods are needed before becoming clinically viable