New insights into the impact of bed rest on lumbopelvic muscles: a computer-vision model approach to measure fat fraction changes

Space agencies plan crewed missions to the Moon and Mars. However, microgravity-induced lumbopelvic deconditioning, characterized by an increased fat fraction (FF) due to reduced physical activity, poses a significant challenge to spine health. This study investigates the spatial distribution of FF in the lumbopelvic muscles to identify the most affected regions by deconditioning, utilizing a computer-vision model and a tile-based approach to assess FF changes. Twenty-four healthy individuals (8 F) were recruited, and automatic segmentation of the lumbopelvic muscles was applied before and after 59 days of head-down tilt bed rest (HDTBR + 59) and 13 days of reconditioning (R + 13). Axial Dixon sequence images were acquired from 3 T magnetic resonance imaging. FF in the lumbar multifidus (LM), lumbar erector spinae (LES), quadratus lumborum, psoas major, gluteus maximus (GMax), gluteus medius (GMed), and gluteus minimus (GMin) muscles from the upper margin of L1 vertebra to the inferior border of GMax muscle were automatically derived using a computer-vision model. Lumbar muscles were segmented into eight tiles (superficial and deep, lateral to medial), and gluteal muscles into regions (anterior/superior for GMed and GMin, superior/inferior for GMax). At HDTBR + 59, the deep centrolateral region at L5/S1 for LM (18.7 ± 15.7%, P < 0.001; d = 0.97) and the deep medial region at Upper L4 for LES (5.4 ± 5.9%, P < 0.001; d = 0.34) showed the largest increase in FF compared with baseline data collection. These regions did not recover at R + 13 (P < 0.05; d ≥ 0.25). These findings highlight the need to target deep fascicles of LM and LES in countermeasure strategies to mitigate microgravity-induced lumbopelvic deconditioning, optimizing spine health, and performance.