Non-specific lower back pain (LBP) is a world-wide public health problem that affects people of all ages. Despite the high prevalence of non-specific LBP and the associated economic burdens, the pathoanatomical mechanisms for the development and course of the condition remain unclear. While intervertebral disc degeneration (IDD) is associated with LBP, there is overlapping occurrence of IDD in symptomatic and asymptomatic individuals, suggesting that degeneration alone cannot identify LBP populations. Previous work has been done trying to relate linear measurements of compression obtained from Magnetic Resonance Imaging (MRI) to pain unsuccessfully. To bridge this gap, we propose to use advanced non-Euclidean statistical shape analysis methods to develop biomarkers that can help identify symptomatic and asymptomatic adults who might be susceptible to standing-induced LBP. We scanned 4 male and 7 female participants who exhibited lower back pain after prolonged standing using an Open Upright MRI. Supine and standing MRIs were obtained for each participant. Patients reported their pain intensity every fifteen minutes within a period of 2 h. Using our proposed geodesic logistic regression, we related the structure of their lower spine to pain and computed a regression model that can delineate lower spine structures using reported pain intensities. These results indicate the feasibility of identifying individuals who may suffer from lower back pain solely based on their spinal anatomy. Our proposed spinal shape analysis methodology have the potential to provide powerful information to the clinicians so they can make better treatment decisions.
Keywords: geodesic logistic regression; intervertebral disc degeneration; longitudinal shape analysis.