Quantitative MRI Assessment of Paraspinal Muscles and Construction of a Predictive Model for Lumbar Disc Herniation

Background Microstructural changes in the paravertebral muscles due to lumbar disc herniation (LDH) remain unclear. This study aims to quantitatively assess paraspinal muscle parameters in LDH patients using MRI and to develop a multivariate imaging-based predictive model. Methods A total of 100 participants were prospectively enrolled, including 53 patients with LDH and 47 healthy controls. This cross-sectional study compared demographic and paraspinal muscle imaging characteristics between groups. All subjects underwent T2 mapping and diffusion tensor imaging (DTI) using a 3.0T MRI scanner. Pain and functional status were assessed using the Numeric Pain Rating Scale (NPRS) and Oswestry Disability Index (ODI). Group comparisons were performed using ANCOVA to adjust for confounders. Partial Spearman correlation evaluated associations between imaging metrics and clinical scores. Stepwise multiple linear regression identified predictors of NPRS and ODI, while logistic regression determined independent LDH risk factors and construction of a nomogram-based predictive model. Results The LDH group had a significantly higher mean age than controls ( P  = 0.023), with no significant differences in other baseline characteristics ( P  > 0.05). ANCOVA revealed lower fractional anisotropy (FA) values in the bilateral erector spinae (ES) at L4/5 and higher apparent diffusion coefficient (ADC) values in the bilateral ES at L4/5, right ES at L5/S1, and bilateral multifidus (MF) in the LDH group ( P  < 0.05 or P  < 0.01). Partial correlation analysis showed that T2 values of specific ES and MF segments were positively associated with ODI and NPRS scores ( P  < 0.05). Multiple regression identified the L4/5 left MF T2 value as the strongest predictor of NPRS and age as the main predictor of ODI ( P  < 0.01). Logistic regression confirmed age, the L3/4 right ES T2 value, and the L5/S1 right ES ADC as independent LDH predictors. The multivariate model yielded the best predictive performance (AUC = 0.772), with nomogram analysis highlighting the L5/S1 right ES ADC as the most influential parameter. Conclusions Paraspinal muscle imaging parameters are effective markers for evaluating muscle status and disease burden in LDH. Multivariate logistic regression identified age, the T2 value of the right erector spinae at L3/4, and the ADC value at L5/S1 as LDH risk factors.