Morphological Analysis of the Human Lower Lumbar Spine Using Sagittal Magnetic Resonance Imaging (MRI) Scans
TimeThursday, April 152:28pm - 2:30pm EDT
DescriptionLinear and planar aspects of the lower lumbar vertebrae, intervertebral discs, and vertebral endplates have already been employed to develop mathematical models and playing a substantial role to characterize the biomechanics of spinal behavior and investigate the potential risk of work-related low back pain. This study proposes a new quantitative exploration of the sagittal morphology of the lower lumbar spine with respect to the disc height and the vertebral body height, along with additional information such as subject height, weight, gender, and age. In this study, sagittal morphological dimensions were measured from two samples of subjects, including 1) archived medical records (AMR) (i.e., MRI scans) collected from 57 subjects and 2) MRI scans collected from 43 subjects with no low back pain history (ASYM). Regions of interest (ROIs) in lower lumbar (i.e. from L3/L4 to L5/S1) were measured using OsiriX© software, including (1) anterior and posterior vertebral body height, (2) anterior and posterior disc height, and (3) cranial and caudal concavity of disc, independently by four raters on a monthly basis to investigate both intra- and inter-rater reliabilities. When it comes to the results of our work, some data doesn’t turn out to be reliable enough because of subjective ROIs and inconsistent image quality. In addition, learning curve as well as both mental and physical fatigue may also influence the measurement reliability. However, it is noteworthy that our study still shows good inter- and intra-reliability overall. Based on statistical analysis, our study reveals that AMR sample had smaller anterior and posterior disc height ratio in disc L5/S1, but greater anterior and posterior vertebral body height ratio in vertebra L3 and L5. Also, ASYM sample had more substantial disc concavity. Through this work, a reliable original reference dataset for future application of machine learning in medical image processing is built, thus helping orthopedic doctors diagnose symptoms.