Regression Models for the Erector Spinae Muscle Mass (ESMM) Cross-Sectional Area: Asymptomatic Populations

Abstract

Understanding low back muscle morphology is critical to understanding spinal loading and the underlying injury mechanisms, which help in characterizing risk and, therefore, minimize low back pain injuries. Individualized erector spinae muscle mass (ESMM) cross-sectional area (CSA) allows biomechanics practitioners to calculate individualized force generating capacities and spinal loadings for given tasks. The objective is to perform morphological analyses and then provide regression models to estimate the ESMM CSA of an individual with his/her subject characteristics. Thirty-five subjects (13 females and 22 males) without low back pain (LBP) history were included in this magnetic resonance imaging (MRI) study. Axial-oblique scans of low back region were used to measure the ESMM CSA. Subject demographics and anthropometrics were obtained and regressed over the ESMM CSA. Best-subset regression analyses were performed. Lean body mass (LBM) and the ankle, wrist, and head indexes were the most frequent predictive variables. Regression models with easy-to-measure variables showed smaller predictive power and increased estimation error compared to other regression models. Practitioners should consider this trade-off between model accuracy and complexity. An individual's ESMM CSA could be estimated by his/her individual characteristics, which enables biomechanical practitioners to estimate individualized low back force capacity and spinal loading.