A Data-Driven Approach to Predict and Classify Epileptic Seizures from Brain-Wide Calcium Imaging Video Data

Area of Science:

• Neuroscience
• Computational Biology
• Epilepsy Research

Background:

• Epileptic seizure prediction remains a critical challenge in epilepsy research.
• Calcium imaging video data captures whole-brain neuronal activity via calcium fluorescence intensity (CFI).

Purpose of the Study:

• To develop a data-driven approach for detecting systemic change-points and predicting epileptic seizures using zebrafish brain-wide calcium imaging data.
• To explore macroscopic patterns differentiating epileptic and control cases for improved classification.

Main Methods:

• An offline training phase involves feature extraction, systemic change-point confirmation, and estimation of system duration ratios.
• An online testing phase utilizes a statistical model to estimate change-points and predict seizure onset.
• Machine learning models and hierarchical clustering are employed for pattern analysis and classification.

Main Results:

• The proposed approach effectively predicts the time range of future epileptic seizures.
• Distinct macroscopic patterns were identified between epileptic and control cases.
• The study proposes a new feature discretization method combined with hierarchical clustering for enhanced visualization and explanation.

Conclusions:

• The developed data-driven method shows promise for accurate epileptic seizure prediction.
• The findings contribute to understanding the macroscopic patterns associated with epilepsy.
• The approach offers a valuable tool for both prediction and analysis in epilepsy research.