Career Center AES Connect

Abstracts

Automatically Detected Spike Ripples Identify the Epileptogenic Zone Better Than Ripples, Fast Ripples, or Hfos in a Multicenter Intracranial Dataset

Abstract number : 1.119
Submission category : 2. Translational Research / 2C. Biomarkers
Year : 2022
Submission ID : 2204253
Source : www.aesnet.org
Presentation date : 12/3/2022 12:00:00 PM
Published date : Nov 22, 2022, 05:24 AM

Authors :
Wen Shi, PhD – Massachusetts General Hospital; Harvard Medical School; Dana Shaw, PhD Candidate – Graduate Program in Neuroscience, Boston University, Boston, Massachusetts, USA; Katherine Walsh, B.S. – Department of Neurology, Massachusetts General Hospital, Boston, Massachusetts, USA; Harvard Medical School, Boston, Massachusetts, USA; Xue Han, PhD – Center for Systems Neuroscience, Boston University, Boston, Massachusetts, USA.; Department of Biomedical Engineering, Boston University, Boston, Massachusetts, USA.; Robert Richardson, MD, PhD – Harvard Medical School, Boston, Massachusetts, USA.; Department of Neurosurgery, Massachusetts General Hospital, Boston, Massachusetts, USA; Uri Eden, PhD – Center for Systems Neuroscience, Boston University, Boston, Massachusetts, USA;Department of Mathematics and Statistics, Boston University, Boston, Massachusetts, USA; Stephen Gliske, PhD – Department of Neurosurgery, University of Nebraska Medical Center, Omaha, Nebraska, USA.; Julia Jacobs, MD – Department of Neuropediatrics and Muscle Disorders, Medical Center-University of Freiburg, Freiburg, Germany;Department of Paediatrics, Cumming School of Medicine, University of Calgary, Calgary, AB, Canada;Department of Neuroscience, Cumming School of Medicine, University of Calgary, Calgary, AB, Canada ;Hotchkiss Brain Institute and Alberta Children’s Hospital Research Institute, University of Calgary, Calgary, AB, Canada; Benjamin Brinkmann, PhD – Bioelectronics Neurophysiology and Engineering Lab, Mayo Clinic, Rochester, Minnesota, USA.; Gregory Worrell, MD, PhD – Bioelectronics Neurophysiology and Engineering Lab, Mayo Clinic, Rochester, Minnesota, USA; William Stacey, MD, PhD – Department of Neurology, University of Michigan, Ann Arbor, MI, USA; Mark Kramer, PhD – Center for Systems Neuroscience, Boston University, Boston, Massachusetts, USA; Department of Mathematics and Statistics, Boston University, Boston, Massachusetts, USA; Catherine Chu, M.D. – Department of Neurology, Massachusetts General Hospital, Boston, Massachusetts, USA.; Harvard Medical School, Boston, Massachusetts, USA.

This abstract has been invited to present during the Translational Research platform session.

Rationale: For patients with focal drug-refractory epilepsy, neurosurgical intervention is the most effective treatment and accurate identification of the pathologic brain tissue responsible for generating the seizures, the epileptogenic zone (EZ), is critical for success. High frequency oscillations (HFOs), including ripples and fast ripples, have been extensively studied as biomarkers for the EZ, but progress has been limited by the presence of physiological HFOs outside of the EZ. Several observations suggest that the co-occurrence of ripples with interictal spikes may be a more specific biomarker for the EZ compared to HFOs alone. Here, we tested this hypothesis on invasive EEG datasets from multiple institutions obtained in patients with focal drug refractory epilepsy.

Methods: We first trained and validated a fully automated combined latent state and convolutional neural network spike ripple detector on 1700 hand-marked events from 18 subjects from 4 institutions using a leave-one-out approach (Shaw et al., AES abstract 2022). We applied the validated detector to intracranial recordings from 20 subjects with Engel 1 outcomes after resective surgery and follow up for ≥1 year (n=6 Freiburg, n=3 McGill University, n=11 Mayo Clinic). In each case, data were collected with >1 kHz sampling rate and referenced to a bipolar montage. SR rates were then computed for each channel. On the same data, we measured ripple, fast ripple, and quantitative HFO rates using current gold standard approaches (Jacobs et al. Neurology, 2018; Gliske et al. Nat. Commun. 2018). For each biomarker, we computed the previously described event rate ratio (Jacobs et al. Ann. Neurol. 2010), where the resected volume was used as the EZ. Thus, numbers closer to 1 indicate higher event rates in the EZ compared to outside of the EZ while numbers closer to -1 indicate the opposite. To test whether spike-ripples better identify the EZ compared to the other biomarkers, we compared the event rate ratios, using a one-way ANOVA followed by paired one-tailed t-tests across all subjects and across each site separately.

Results: Our preliminary dataset included 20 subjects with drug refractory epilepsy due to several etiologies (11F, mean age 27; range, 8-62). The median duration of data analyzed was 60 min/subject (range, 10-120 min) and the median channel count was 90/subject (range 35-131). Across all subjects, there was a difference in event rate ratio between groups (p=4e-3), where the spike ripple event ratio was significantly higher than each of the other biomarkers’ event rate ratios (p < 0.05). These results persisted in subgroup analysis in two of the three centers. In the latter case, 3 outliers were detected; two of whom had spike ripples detected in a non-resected hippocampus and one of whom had very sparse detections.
Translational Research