کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

Resting-State functional magnetic resonance imaging Biomarkers for Predicting Treatment Response in Major Depressive Disorder

نویسندگان :

Zahra Nakhaei¹ Gholam-Ali Hossein-Zadeh²

1- دانشگاه تهران 2- دانشگاه تهران

کلمات کلیدی :

Major depressive disorder،Functional connectivity،Treatment outcome،Regression Analysis

چکیده :

Background: Major depressive disorder (MDD) is a common psychiatric illness with heterogeneous treatment outcomes, often requiring trial-and-error approaches for treatment. The development of neuroimaging-based biomarkers could help design more effective and personalized interventions. Method: In this study, resting-state fMRI data from six patients with MDD from OpenNeuro database were used. ROI-to-ROI functional connectivity matrices were calculated. To identify predictive features for the effect of cognitive-behavioral therapy (CBT), a random forest algorithm was used and three significant connections were selected. These features were then used as input to a support vector regression (SVR) model. The model was evaluated using leave-one-out cross-validation. Result: Out of 6670 possible connections, three connections were selected which were correlated with clinical improvement of CBT including: i) Right fusiform gyrus with right superior temporal gyrus, ii) Right insula with right cerebellar lobule 3 iii) Right anterior cingulate gyrus with left angular gyrus, The optimized SVR model achieved a mean square error (MSE) of 5.30 and a coefficient of determination (R²) of 0.23. Conclusion: These results suggest that a small, interpretable set of brain connections can be used to predict symptom severity after treatment in patients with MDD. The findings highlight the potential of resting-state fMRI for use in personalized treatment planning.