Traditionally paraspinal muscle quality has been assessed using morphological characteristics and measures of macroscopic composition such as muscle fat infiltration on computer tomography (CT) and magnetic resonance imaging (MRI). The evidence for associations between these characteristics and low back pain (LBP) is conflicting. We sought to investigate the role of non-morphological quantitative parameters of muscle quality (primarily those derived from multi-echo acquisitions) in low back pain.
We present here a pipeline for automatic quantification of morphological characteristics, segmentation of intra- and extra-muscular fat, and the calculation of quantitative T2 and T2* relaxation times and Decay Variance. We have applied this processing pipeline to a prospective cohort study of 20 patients aged 29-74 who experienced cryptogenic LBP, with no previous spinal surgery and meeting the NIH definition for chronicity. Patients had lumbar MRI with standardised acquisition parameters at a single centre. Data was collected prospectively. Due to the semi-automated nature of the processing pipeline intra- and inter-rater reliability was high for all measures. Future work will focus on the integration of DTI parameters into the pipeline, and correlation with patient clinical features.