Parametric subtracted post-ictal diffusion tensor imaging for guiding direct neurostimulation therapy

Parametric subtracted post-ictal diffusion tensor imaging (pspiDTI) is a novel imaging technique developed at our center to visualize transient, patient-specific, ictal-associated water diffusion abnormalities in hippocampal-associated axonal tissue. PspiDTI can elucidate putative connectivity patterns, tracing ictal propagation following a partial-onset seizure without generalization secondarily. PspiDTI compares two DTI volumes acquired during the early post-ictal period (<4 hr), and baseline inter-ictal interval (>24 hr post-seizure). This technique performs a voxel-wise parametric test to identify statistically significant transient ictal-associated changes in water diffusivity involving white matter (WM). Our technique was applied to six patients with refractory partial-onset epilepsy who were candidates for direct cortical responsive neurostimulation (RNS) therapy. Global and region-specific fractional anisotropy decreases, relative to baseline, were detected in all patients with a 17.01% (p < .01) relative mean decrement, while trace increases were found in 6/6 (100%) patients with a 13.30% (p < .01) relative global mean increment. Changes in diffusivity were anatomically compared with transient hyper-perfusion as detected by subtracted ictal SPECT co-registered to MRI (SISCOM). In 5/6 (83.33%) patients, alterations in WM diffusivity were detected adjacent to the SISCOM signal localized predominantly in gray matter. In 4/6 patients, post-implant RNS electrocorticography revealed early ictal propagation between implanted RNS depth leads guided by pspiDTI, hence validating concordant abnormal diffusivity regions detected by our technique. PspiDTI can complement the conventional pre-surgical evaluation to provide additional crucial information regarding WM ictal-propagation pathways between predominantly gray matter ictal-onset zones. When incorporated into a multi-modality pre-surgical workflow, pspiDTI can aid in defining critical nodes between ictogenic regions. This information can be used to strategically implant a limited set of two RNS depth leads for maximizing the extent to which direct cortical RNS can modulate a potentially extensive epileptogenic network.