Development of 3D method to assess intramuscular spatial distribution of fat infiltration in patients with rotator cuff tear: reliability and concurrent validity

BACKGROUND: Intramuscular fat infiltration is a critical factor in surgical decision-making and is the most important factor used to prognosticate surgical repair outcomes in patients with rotator cuff tears. Quantitative 3D assessment of total rotator cuff fat infiltration in patients with rotator cuff tears has been realized. However, a reproducible method to evaluate 3D spatial distribution of rotator cuff intramuscular fat has not been established. The objective of this study was to establish the reproducibility, change detectable beyond error, and concurrent validity of a semi-automated method to evaluate the 3D spatial distribution of fat infiltration and muscle volume in patients with rotator cuff tears. METHODS: Thirteen consecutive patients diagnosed with symptomatic rotator cuff pathology and 3.0 T MRI confirmation at a single center were included. Fat-water imaging was used to quantify 3D intramuscular fat (%fat) in sagittal oblique sequences and intramuscular spatial distribution with the semi-automated technique. Each rotator cuff muscle was manually segmented yielding %fat in four axial intramuscular quartile-regions (superior-inferior; Q1–4) and three sagittal (medial/ intermediate/ lateral) regions. Reliability and concurrent validity of %fat and whole muscle volume were calculated with intraclass correlation coefficients (ICC). RESULTS: Intra-rater reliability for intramuscular sagittal divisions (ICC = 0.93–0.99) and axial divisions (ICC = 0.78–0.99) was good/excellent. Inter-rater reliability for %fat (ICC = 0.82–0.99) and volume (ICC = 0.92–0.99) was good/excellent. Concurrent validity with commercialized software showed good/excellent agreement (ICC = 0.66–0.99). CONCLUSIONS: A new semi-automated method to assess 3-dimensional intramuscular distribution of fat infiltration in patients with rotator cuff tears using advanced MR imaging demonstrates high intra and inter-rater reliability and good concurrent validity. Minimal detectable change thresholds established facilitate clinical interpretation for future clinical application of this technique to assess change and treatment efficacy in patients with rotator cuff tears.