Prediction of the Erector Spinae Muscle Lever Arm Distance for Biomechanical Models
Type of Degreedissertation
Industrial and Systems Engineering
MetadataShow full item record
Low Back Pain (LBP) remains the U.S.’s most significant and costly injury. Improved biomechanical modeling of the lumbar spine may allow better evaluation of LBP risk. To calculate the forces acting on the spine, accurate biomechanical model inputs are required. However, some biomechanical model inputs are limited by assumptions. One of the most vital model inputs, the mechanical lever arm of the erector spinae muscle mass, (ESMLA), is typically approximated using a fixed value (5 cm or 2 inch) to simplify biomechanical models. This assumption decreases the sensitivity and applicability of models as well as their credibility. The objective of this study was to develop regression models to estimate the ESMLA distance based solely on (1) easily measured subject variables (gender, age, height, and weight) and (2) some additional anthropometric variables (i.e., lean body mass, sitting height, shoulder width). This will allow currently available biomechanical models to incorporate subject specific parameters and should improve model predictions and risk estimations. In addition to the ESMLA distances at several inter-vertebral disc levels in the lower lumbar region, other morphological parameters of musculoskeletal structure such as the cross-sectional areas (CSA) of the erector spine muscle mass (ESMM) and the inter-vertebral disc (IVD) are investigated. Regression models were also developed for the CSA of the ESMMs at each IVD level. Magnetic Resonance Images (MRI) were used in this study. They were obtained from (1) a historical data base and (2) a newly conducted study at the Auburn MRI Research Center. The ESMLA distances and the CSA measurements were measured from axial oblique MRI scans by using architectural design software. Measurements were then statistically investigated to determine the relationships between the measurement and subject variables (characteristic and anthropometrics). Results indicate that the ESMLA distance and ESMM size can be easily and reliably estimated using subject variables. The results of the present study found that using a fixed ESMLA value could cause errors be as great as 20%. The average error percentage of using the fixed value was 8%. Using an empirically derived average value for a IVD level and gender could cause approximately 5% error in ESMLA distances. On the other hand, using regression models suggested in the present study yielded smaller error percentages. For example, the average error was approximately 4.3% for regression models that had easy to measure anthropometric variables (i.e., height and weight). Regression models that had more predictive variables (i.e., ankle, wrist, and knee indexes), however, can provide much smaller prediction errors. The average absolute residual percentage was 2.15% for the L3/L4 level, 2.39% for the L4/L5 level, and 3.67% for the L5/S1 level. The advantage of using regression equations is that smaller prediction errors in ESMLA distances result in smaller error in spinal loading calculations, especially for extreme subjects.