Auto-tracking system for human lumbar motion analysis

Article type: Research Article

Authors: Sui, Fuge | Zhang, Da | Lam, Shing Chun Benny | Zhao, Lifeng | Wang, Dongjun | Bi, Zhenggang | Hu, Yong

Affiliations: Department of Orthopaedic Surgery, Longnan Hospital, Daqing, Heilong Jiang, China | Department of Orthopaedics and Traumatology, The University of Hong Kong, Hong Kong, China | Daqing Youtian Gerneral Hospital aqing, Heilong Jiang, China | Department of Radiology, University of Pennsylvania, Philadelphia, PA, USA | Department of Orthopaedic surgery, The First Affiliated Hospital of Harbin Medical University, Harbin, China

Note: [] Corresponding author: Dr. Y. Hu, 12 Sandy Bay Road, Hong Kong, Duchess of Kent Children's Hospital, Department of Orthopaedics Surgery, The University of Hong Kong, Pokfulam, Hong Kong, China. Tel.: +852 297 40359; Fax: +852 297 40335; E-mail: yhud@hkusua.hku.hk

Abstract: Previous lumbar motion analyses suggest the usefulness of quantitatively characterizing spine motion. However, the application of such measurements is still limited by the lack of user-friendly automatic spine motion analysis systems. This paper describes an automatic analysis system to measure lumbar spine disorders that consists of a spine motion guidance device, an X-ray imaging modality to acquire digitized video fluoroscopy (DVF) sequences and an automated tracking module with a graphical user interface (GUI). DVF sequences of the lumbar spine are recorded during flexion-extension under a guidance device. The automatic tracking software utilizing a particle filter locates the vertebra-of-interest in every frame of the sequence, and the tracking result is displayed on the GUI. Kinematic parameters are also extracted from the tracking results for motion analysis. We observed that, in a bone model test, the maximum fiducial error was 3.7%, and the maximum repeatability error in translation and rotation was 1.2% and 2.6%, respectively. In our simulated DVF sequence study, the automatic tracking was not successful when the noise intensity was greater than 0.50. In a noisy situation, the maximal difference was 1.3 mm in translation and 1° in the rotation angle. The errors were calculated in translation (fiducial error: 2.4%, repeatability error: 0.5%) and in the rotation angle (fiducial error: 1.0%, repeatability error: 0.7%). However, the automatic tracking software could successfully track simulated sequences contaminated by noise at a density ⩽ 0.5 with very high accuracy, providing good reliability and robustness. A clinical trial with 10 healthy subjects and 2 lumbar spondylolisthesis patients were enrolled in this study. The measurement with auto-tacking of DVF provided some information not seen in the conventional X-ray. The results proposed the potential use of the proposed system for clinical applications.

Keywords: Auto-tracking, digitized video fluoroscopy, spine motion, particle filter, lumbar spine, vertebral body

DOI: 10.3233/XST-2011-0287

Journal: Journal of X-Ray Science and Technology, vol. 19, no. 2, pp. 205-218, 2011