Using DXA to measure body composition is a solid choice, since it gives reasonably precise estimates of lean mass distribution compared to simpler methods like BIA.
What caught my attention more is the dietary control part. Self-reported dietary logs, even with photos and interviews, can still introduce variability, especially over a 16-week intervention.
In studies trying to link exercise with skin-related outcomes, dietary factors (protein intake, micronutrients, total energy balance) can be a pretty significant confounder. It would be interesting to see how tightly those variables were controlled or normalized in the analysis.