Preoperative structural networks based on DTI predicts initial subthalamic nucleus stimulation outcome in parkinson's disease.

Deep brain stimulation (DBS) is widely used to treat Parkinson's disease (PD), but its efficacy varies. This study aimed to investigate how preoperative structural network influences subthalamic nucleus deep brain stimulation (STN-DBS) outcome in Parkinson's disease (PD). This study retrospectively collected 93 patients, and divided them into a low improvement group (LIG) and a high improvement group (HIG). Preoperative structural networks were constructed from diffusion tensor images using probability constrained spherical deconvolution algorithm. This study compared topological characteristics between groups and explored the prognostic value of structural networks. Compared to HIG, LIG has a longer normalized characteristic path length and diminished inter-regional connections within left frontal lobe. Normalized characteristic path length is negatively correlated with DBS outcome, while connection strength is positively correlated with DBS outcome. Notably, the fusion method of clinical phenotype and network characteristics has better predictive power for postoperative DBS outcome than either the clinical method or the network method. This study reveals that both normalized characteristic path length and connectivity between the left superior frontal gyrus (central region) and the left medial frontal gyrus (ventral lateral region) are associated with initial DBS efficacy, which could be preoperative biomarkers of DBS outcome.