Volumetric Analysis of the Deltoid Muscle From Neck-to-Knee Dixon MRI: A pilot population assessment of a posterior intramuscular injection target

Abstract Background: Deltoid muscle volume varies across anatomical regions and population groups, with potential implications for intramuscular injection site selection. Evidence comparing middle and posterior deltoid volumes in vaccine-eligible populations is limited. Purpose: To compare middle and posterior deltoid muscle volumes by vaccine eligibility group, sex, and side. Methods: Deltoid muscles were manually segmented into anterior, middle and posterior regions in 43 UK Biobank Dixon MRI scans. A MONAI deep learning model trained on the annotations (Dice score = 0.78) segmented deltoid muscles of 5,239 participants, categorised by UK flu vaccination eligibility: A (≥65 years without risk conditions, n=1,760), B (≥65 years with risk conditions, n=1,744), and C (<65 with risk conditions, n=1,735). Bilateral-average posterior and middle volumes were compared using a paired t-test. A linear mixed-effects model assessed posterior-middle differences by group, sex and side. Results: Mean age was 66.7 ± 7.2 years and body mass index was 27.5 ± 5.7 kg/m2 (mean ± SD). Posterior deltoid volume exceeded the middle deltoid volume by 55.36 ml (95% CI, 54.14-56.58, p<0.0001). Posterior-middle differences were largest in Group A, followed by B and then C, greater in men, and greater on the right side (all p<0.0001). Mean segmentation-quality score was 4.27 ± 0.73 on a five-point Likert scale and weighted inter-observer kappa was 0.583 (95% CI, 0.452 - 0.713). Conclusions: Among individuals eligible for the UK flu immunisation, posterior deltoid volume was greater than middle deltoid volume. These findings support further anatomical assessment of the posterior deltoid as a potentially safer injection target.