Abstracts

Rationale: Subtraction ictal single-photon emission computed tomography (SPECT) co-registered to MRI (SISCOM) is a valuable non-invasive neuroimaging tool in the presurgical evaluation of refractory focal epilepsy. The complex perfusion patterns observed in SISCOM map may reflect the ictal onset and propagation pathways. The aim of this study is to understand the dynamical changes in connectivity within ictal perfusion patterns during the baseline, pre-ictal, ictal and post-ictal time periods in a consecutive series of patients who underwent ictal and interictal SPECT as part of their presurgical evaluation and later underwent stereo-EEG (SEEG) evaluation between 01/15 and 12/16. Methods: To evaluate the relationship between SISCOM and SEEG seizures, seizures with similar semiological changes during ictal SPECT and SEEG were selected. For every patient, a 2 minutes baseline segment of interictal SEEG, at least 2 hours preceding a seizure, was identified. Patients were excluded from the analysis if SPECT injection was undertaken during the postictal phase, poor correlation between semiology during ictal SPECT and SEEG seizures, non-availability of raw SEEG baseline and ictal episodes and poor quality of SEEG data. The SISCOM map was co-registered to the SEEG implantation map and the SISCOM z-score for each SEEG contact was extracted. Seizures were segmented into onset, middle and offset segments by dividing the total ictal time period into 3 segments of equal duration. Interaction between hyperperfused, hypoperfused, and baseline perfused brain regions were quantified using directed transfer function. Differences in connectivity between different perfusion patterns were quantified by estimating the t-statistics. Relationship between perfusion score and connectivity was quantified using correlation analysis. Finally, connectivity between hyperperfused electrode contacts in- and outside the resected brain area of surgically treated patients was determined. Correction for multiple comparisons was performed using false discovery rate. Results: A total of 70 patients met the specific inclusion criteria. In 51 patients, an adequate number of hyper, hypo and baseline perfused contacts were available for analysis. Statistically significant and higher information flow from hyper- to hypoperfused and baseline to hypo-perfused brain regions were observed during the pre-ictal and onset segments across all frequency bands (Fig 1A). A statistically significant correlation was observed between the perfusion z-score and the total information outflow from each brain region across all frequency bands (Fig 1B). Finally, in patients who were seizure free after surgery (N=12) there is a statistically significant and higher information flow from hyperperfused and baseline perfused resected SEEG contacts to the corresponding non-resected resected SEEG contacts in the ß-? frequency bands (Fig 1C) while no such relationship was observed across non-seizure free patients (N=6) (Fig 1D). Conclusions: The principal findings of this study are that regions with higher perfusion score tend to generate information during ictal phase of a seizure. Information flows through the gradient of the perfusion z-score, i.e. regions with higher perfusion z-score tend to drive regions with lower perfusion score. Finally, the study finds that connectivity between resected and non-resected hyperperfused tissue is different in patients who are seizure free after resective surgery. The results from the study imply that connectivity analysis of SEEG can be used to identify critical nodes within the hyperperfusion patterns involved in seizure onset and propagation. Funding: Not applicable