Abstract
Objective
Quantification of the epileptogenic zone network (EZN) most frequently implies analysis of the seizure onset. However, important information can also be obtained from the postictal period, characterized by prominent changes in the EZN. We used permutation entropy (PE), a measure of signal complexity, to analyze the peri-ictal stereoelectroencephalography (SEEG) signal changes with emphasis on the postictal state. We sought to determine the best PE-derived parameter (PEDP) for identifying the EZN.
Methods
Several PEDPs were retrospectively computed on SEEG-recorded seizures of 86 patients operated on for drug-resistant epilepsy: mean baseline preictal entropy, minimum ictal entropy, maximum postictal entropy, the ratio between the maximum postictal and the minimum ictal entropy; the ratio between the maximum postictal and the baseline preictal entropy. The performance of each biomarker was assessed by comparing the identified epileptogenic contacts or brain regions against the EZN defined by clinical analysis incorporating the Epileptogenicity Index and the Connectivity Epileptogenicity Index methods (EZNc), using the Receiver Operating Characteristic and Precision-Recall.
Results
The ratio between the maximum postictal and the minimum ictal entropy (defined as the Permutation Entropy Index, PEI) proved to be the best-performing PEDP to identify the EZNC. It demonstrated the highest Area Under the Curve (AUC) and F1 score at the contact level (AUC 0.72; F1 0.39) and at the region level (AUC 0.78; F1 0.47). PEI values gradually decreased between the EZN, the propagation network and the non-involved regions. PEI showed higher performance in patients with slow seizure-onset patterns than in those with fast seizure-onset patterns. The percentage of resected epileptogenic regions defined by PEI was significantly correlated with surgical outcome.
Significance
PEI is a promising tool to improve the delineation of the EZN. PEI combines ease and robustness in a routine clinical setting with high sensitivity for seizures without fast activity at seizure onset.
DIC