Towards Data-Driven Tractography for Surgical Planning

Motivation: For brain tumor patients, surgeons must identify essential language regions surrounding the tumor to minimize neurological morbidity. Goal(s): We aim to develop data-driven approaches for preoperatively identifying the peritumoral regions that are essential for sustaining language function. Approach: This study included 71 surgical glioma patients who underwent preoperative DTI, preoperative cortical language mappings by transcranial magnetic stimulation (TMS) and functional MRI (fMRI), and perioperative language evaluations. Tractography was performed using cortical functional data and normalized diffusion metrics. Results: Resecting tracts localized by TMS cortical language sites at a particular normalized fractional anisotropic threshold strongly predicted aphasic deficits. Impact: These findings outline a data-driven, patient-specific approach for using tractography to localize the language-essential regions surrounding the tumor.