Key result
Automated FWHT-based neural network achieves ~99% accuracy classifying focal and non-focal EEG signals.
Why the study?
Visual examination of long EEG signals to discriminate focal from non-focal classes during neurosurgery is time-consuming and prone to error.
Population
EEG signals from University of Bonn and Bern-Barcelona datasets, including 3750 pairs of NFC and FC signals
Comparison
Automated FWHT, entropy features, and ANN classification vs existing techniques
Design
Algorithm development and validation study with 10-fold cross-validation
Authors
Loading...
May aid epileptogenic zone localization; leaves open prospective clinical validation before adoption.
An automated FWHT and ANN-based approach provides highly accurate discrimination of focal and non-focal EEG signals, potentially aiding in the localization of the epileptogenic zone.
Prasanna et al. (2020) studied Epilepsy. Automated diagnosis using Fast Walsh-Hadamard Transform (FWHT) and Artificial Neural Network (ANN) vs. Conventional diagnosis method / existing techniques was evaluated on Classification accuracy of non-focal and focal EEG signals. An automated diagnosis method using Fast Walsh-Hadamard Transform and an artificial neural network achieved 99.50% accuracy in classifying focal and non-focal EEG signals.
Synapse has enriched one closely related paper. Consider it for comparative context: