A Framework for Global Background Subtraction in Fiber Diffraction Patterns from Striated Muscle

Accurate background subtraction is critical in the analysis of X-ray fiber diffraction patterns from striated muscle. Diffuse scattering and low signal-to-noise ratio complicate the separation of signal from background. Existing methods are applied heuristically, without guidance on parameter selection, leading to variability in following analyses. We propose a framework for background removal in muscle fiber diffraction images that evaluates methods and selects optimal parameters. It combines quantitative metrics and validation using synthetic features to guide the optimization. Minimizing the aggregate loss produced visually plausible results in the representative cases examined by the authors; independent blinded validation was not performed. The loss metric and its components allow currently used approaches to be compared within a single reproducible scheme. Individual metrics show trade-offs among approaches, such as preservation of detail or oversubtraction. Additionally, we evaluate parametric fitting of the equatorial streak and isotropic general background, which may be combined with the optimization framework. The framework improves the analysis of fiber diffraction patterns by providing a reproducible quantitative basis for the choice of background subtraction method and its parameters.