Estimation of the epileptogenic-zone with HFO sub-groups exhibiting various levels of epileptogenicity
| Autor: | Matthias Dümpelmann, Daniel Lachner-Piza, Thomas Stieglitz, Andreas Schulze-Bonhage, Julia Jacobs |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
0301 basic medicine
Physics Brain Mapping Epilepsy business.industry Pattern recognition Brain waves Epileptogenic zone Brain Waves Intracranial eeg Coincidence Hafnium oxide 03 medical and health sciences 030104 developmental biology 0302 clinical medicine Humans Electrocorticography Artificial intelligence Deterministic method business Cluster analysis 030217 neurology & neurosurgery |
| Zdroj: | EMBC |
| Popis: | High-Frequency-Oscillations (HFO) are biomarkers of the epileptogenic-zone (EZ) and thus a potential aid in guiding epilepsy-surgery. HFO are normally sub-divided according to their oscillating-frequency into Ripples (80-250 Hz) and Fast-Ripples (FR) (250-500 Hz) and are known to also occur in the non-epileptic brain. We address two challenges faced by HFO: firstly, estimating the margins of the EZ using the HFO occurrence-rate from each intracranial EEG channel; secondly, selecting HFO sub-groups with a higher probability of being purely epileptic. We propose the clustering of channels with high HFO occurrence-rates as a deterministic method to delimit the EZ. Additionally, we perform the EZ estimation using 9 sub-groups of HFO; these sub-groups are determined by their temporal and spatial coincidence with intracranial interictal-epileptic-spikes (IES) and are assumed to have varying levels of epileptogenicity. The EZ estimated with the different HFO-sub-groups are compared between themselves and with a proxy of the factually undefinable EZ, namely the resected-volume (RV). The proposed clustering method proved to be deterministic and allowed estimating the EZ for each patient and each HFO-sub-group. Those Ripples assumed to be more epileptogenic occurred in lower numbers than all Ripples but showed the highest correspondence with the RV. All FR sub-groups showed a high specificity to the RV. The proposed clustering method successfully extracted the information from the HFO occurrence-rate to estimate the EZ. The selection of more epileptogenic HFO based on their coincidence with IES proved to be of value for both Ripples and FR. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|