Deep Learning Tool for Analyzing EEG Images in Epilepsy
Author Information
Author(s): Makaram Navaneethakrishna, Gupta Sarvagya, Pesce Matthew, Bolton Jeffrey, Stone Scellig, Haehn Daniel, Pomplun Marc, Papadelis Christos, Pearl Phillip, Rotenberg Alexander, Grant Patricia Ellen, Tamilia Eleonora
Primary Institution: Boston Children’s Hospital, Harvard Medical School
Hypothesis
Electrodes recording from the epileptogenic brain tissue have lower TF visual complexity compared to others.
Conclusion
The study presents a computerized tool that accurately localizes the epileptogenic zone in the brain using brief interictal EEG data.
Supporting Evidence
- Contacts inside the seizure onset zone exhibit lower UAE than outside.
- Larger differences in deep layers indicate significant findings.
- Machine learning algorithms improved localization accuracy to 7 mm.
- Study provides a novel approach to localize the epileptogenic zone.
- Results suggest the method could enhance surgical planning for epilepsy.
Takeaway
This study uses a computer program to help doctors find the part of the brain causing seizures by looking at brain wave pictures.
Methodology
Analyzed interictal EEG data from 20 children, created time-frequency images, and used a deep learning model to assess visual complexity.
Potential Biases
Potential bias due to the retrospective nature of the study.
Limitations
Limited sample size and heterogeneity of the pediatric cohort.
Participant Demographics
20 pediatric patients, mean age 12 years, 6 females.
Statistical Information
P-Value
0.00016
Confidence Interval
1.18–0.136
Statistical Significance
p < 0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website