Use of machine learning in predicting the efficacy of repetitive transcranial magnetic stimulation on treating depression based on functional and structural thalamo-prefrontal connectivity: A pilot study

Abstract

Repetitive transcranial magnetic stimulation (rTMS) is a noninvasive, safe, and efficacious treatment for major depressive disorder (MDD). However, the antidepressant efficacy of rTMS greatly varies across individual patients. Thus, markers that can be used to predict the outcome of rTMS treatment at the individual level must be identified. Thalamo-cortical connectivity was abnormal in patients with MDD, and was normalized after rTMS treatment. In the current study, we investigated whether the resting-state functional and structural thalamo-cortical connectivity could be utilized to predict the rTMS treatment efficacy by employing support vector machine regression analysis. Results showed that the Hamilton Depression Scale scores of patients with MDD decreased after rTMS treatment. The functional connectivity of mediodorsal nucleus with prefrontal cortex predicted the rTMS treatment improvement, whereas the functional connectivity of other thalamic nuclei with cerebral cortex did not predict the treatment efficacy. The brain areas that contributed the most to the prediction were dorsal lateral prefrontal cortex, ventral lateral, and orbital and medial prefrontal areas. The improvement in the outcome of rTMS treatment could also be predicted by the thalamo-prefrontal structural connectivity. No statistically significantly difference in thalamo-cortical connectivity was observed between early improvers and early non-improvers. These results suggested that the thalamo-prefrontal connectivity can predict the rTMS treatment improvement. This study highlighted the crucial role of the thalamo-prefrontal connectivity as a neuroimaging marker in the treatment of depression via rTMS.