| Title: |
Conference proceeding: Patient-Specific Deep Learning for IED Detection in tDCS Epilepsy Treatment: A Synthetic Data Augmented Approach |
| Authors: |
Peuvrier, M; Kayabas, M, A; Köksal-Ersöz, E; Daoud, M; Bartolomei, F; Merlet, I; Wendling, F |
| Contributors: |
Laboratoire Traitement du Signal et de l'Image (LTSI); Université de Rennes (UR)-Institut National de la Santé et de la Recherche Médicale (INSERM); Centre Inria de Lyon; Institut National de Recherche en Informatique et en Automatique (Inria); Computation, Cognition and Neurophysiology (CRNL-COPHY); Centre de recherche en neurosciences de Lyon - Lyon Neuroscience Research Center (CRNL); Université Claude Bernard Lyon 1 (UCBL); Université de Lyon-Université de Lyon-Université Jean Monnet - Saint-Étienne (UJM)-Institut National de la Santé et de la Recherche Médicale (INSERM)-Centre National de la Recherche Scientifique (CNRS)-Université Claude Bernard Lyon 1 (UCBL); Université de Lyon-Université de Lyon-Université Jean Monnet - Saint-Étienne (UJM)-Institut National de la Santé et de la Recherche Médicale (INSERM)-Centre National de la Recherche Scientifique (CNRS); Assistance Publique - Hôpitaux de Marseille (APHM) |
| Source: |
BioEM ; https://hal.science/hal-05393784 ; BioEM, Jun 2025, Rennes, France |
| Publisher Information: |
CCSD |
| Publication Year: |
2025 |
| Collection: |
Université Jean Monnet – Saint-Etienne: HAL |
| Subject Terms: |
Machine learning; patient specific; epilepsy; detection; synthetic data; [INFO]Computer Science [cs]; [SDV]Life Sciences [q-bio] |
| Subject Geographic: |
Rennes; France |
| Description: |
International audience ; Transcranial direct current stimulation (tDCS) shows promise for drug-resistant epilepsy patients, but monitoring treatment efficacy remains challenging. While current monitoring relies on subjective seizure diaries, electroencephalogram (EEG) recordings offer a more objective approach through the detection of interictal epileptiform discharges (IEDs). However, automated IED detection faces challenges with varying recording conditions and treatment-induced changes in signal patterns. We present a novel framework that enhances patient-specific deep learning models with synthetically generated EEG data for automated IED detection. Our approach incorporates personalized simulations of both the patient's epileptic activity and their tDCS treatment response. We expect to show that this synthetic data augmentation improves model resilience to recording variations and maintains consistent performance across treatment sessions. This method should reduce the reliance on expert annotation while providing robust, objective monitoring of tDCS treatment outcomes. |
| Document Type: |
conference object |
| Language: |
English |
| Availability: |
https://hal.science/hal-05393784; https://hal.science/hal-05393784v1/document; https://hal.science/hal-05393784v1/file/Abstract_bioEM_v0.pdf |
| Rights: |
info:eu-repo/semantics/OpenAccess |
| Accession Number: |
edsbas.9F00DD89 |
| Database: |
BASE |