Magnetic Resonance Imaging Textural Changes Are More Sensitive Than Volumetric Changes in the Amygdala of Cocaine Use Disorder Patients

MRI morphometry tracks disease progression by measuring volumetric changes in brain structures, but it could miss subtle microstructural alterations. Textural changes have been shown to detect these microstructural changes and quantify gray-level spatial patterns more sensitively than volumetry. Similar approaches have detected early hippocampal changes in Alzheimer's disease, suggesting textural analysis could uncover clinically relevant biomarkers in other disorders. In this work, we analyzed 25 participants with cocaine use disorder undergoing repetitive transcranial magnetic stimulation (rTMS) from the SUDMEX-TMS dataset, comparing amygdala volumetric and textural changes over time. Given the amygdala's established role in addiction, we hypothesized that textural features would provide greater sensitivity to rTMS-induced changes than volumetry alone. Feature selection for left and right amygdala was performed using manual selection and LASSO regression with subject-level cross-validation, followed by variance inflation factor (VIF) correction and linear mixed-effects model (LMEM) analysis. The left amygdala model retained 8 of 14 features (4 after VIF correction), achieving an ICC of 0.74, while the right model retained 6 features (4 after VIF correction) with an ICC of 0.59, indicating hemispheric asymmetry in structural stability. Temporal sensitivity analysis revealed that VIF-corrected features showed 2–8 folds higher temporal sensitivity than volume. Five features were shared across hemispheres (Jaccard similarity = 0.56), suggesting partially bilateral but lateralized microstructural determinants of amygdala volume. These GLCM-derived texture features represent potential markers for rTMS response in CUD. Future studies will investigate the mapping of these features to underlying neuropathology.