Quantum Application for EEG Anomaly Detection

Rajiv Rajiv • 4/7/2026
Share:

The Challenge

Epilepsy onset detection is slow when using classical ML techniques. How to speed this up while also improving accuracy and precision?

Our Solution

By leveraging quanvolutional neural networks, we developed a novel approach that uses DuoQ-EpiNet: a dual-track hybrid framework that integrates quantum and classical deep learning models for robust seizure classification using the University of Bonn EEG dataset. In the first track, handcrafted statistical and spectral descriptors are extracted from the raw EEG signals and subsequently analyzed using a 1D Convolutional Neural Network (CNN) to learn discriminative temporal representations. In parallel, the second track with wavelet approach transforms the EEG signals into scalogram images, which are processed through a Hybrid Quanvolutional Classical Convolutional Neural Network (HQCNN) equipped with a Fixed Quantum Filter Circuit to generate expressive quantum feature maps followed by classical CNN. The latent representations obtained from both tracks are then fused and passed through fully connected layers to perform the final binary classification. We published our research in EPJ Quantum Technology recently with the results of this approach and are in the process of commercializing it via pilots with interested sponsoring healthcare providers and neurology labs researching epilepsy. The research paper can be found in EPJ Quantum Technology. Please schedule a meeting if interested in discussing further details and engaging with Pivotport in a pilot project.

Results

Systematic comparison of the proposed DuoQ-EpiNet model by tweaking quantum hyperparameter based variants, state-of-the-art HQCNN architectures, as well as the best classical transfer learning models have demonstrated that the proposed model performs better than all evaluated variants. Among all evaluated configurations, the proposed DuoQ-EpiNet Binary Dual-Track (P-B-D) model achieved outstanding performance of 98.50% accuracy with its FQFC employed in Track 2 contrived with quantum hyperparameters of n_shots = 1000 and n_layers = 1 . Performance in data-scale studies ranging from 5% to 100% shows that DuoQ-EpiNet outperforms traditional baselines. Its generalization ability is confirmed by evaluation on the CHB-MIT scalp EEG dataset. The model maintains its stability at low noise densities with only slight performance deterioration, according to NISQ robustness study employing density matrix simulations with depolarizing, amplitude damping, phase damping, and readout noise.

Future Applications

In this study, we have formulated a dual-track hybrid approach named DuoQ-EpiNet, which incorporates statistical and spectral features of EEG signals and multichannel quantum feature map of scalogram to classify epileptic seizures in EEG signals. The proposed approach, which involves a structured fixed quantum filter circuit in the fixed Quanvolutional phase implemented in Quantum simulator owing to cost constraints, is capable of obtaining highly discriminative feature maps than other existing fixed quantum filter circuit approaches. The performance of the DuoQ-EpiNet approach, as demonstrated with a variety of models and datasets, showcases that it performs better than the conventional models in scenarios where deep learning models fail specifically in low-data regime. This validation focuses on the intricacies of our approach as it encompasses significant complexity in its formulation, which whenever used in a hospital setting, are dependent on the sparse EEG signals they can access. The proposed model uses a frugal 4-qubit quantum circuit with structured encoding layer and shalow circuit depth satisfying the NISQ constraints. For offline EEG analysis, the execution time from a JAX framework stays within reasonable bounds. Besides promoting methodological innovation, another significant aspect of the proposed end-end pipeline is its effort to fill the gap existing between classical machine learning approaches and quantum paradigms regarding healthcare AI where data scarecity exists. The ability of DuoQ-EpiNet to generalize well on previously tes data underscores its potential as a clinically valuable tool for reliable seizure detection in real-world settings. In future work, we aim to extend this architecture to larger, multi-institutional datasets to enhance clinical trust and on latest noisy intermediate-scale quantum technology for real time implementation Epileptic seizure prediction. Meanwhile, the proposed method may also find its potential implementation within the context of ambulatory EEG monitoring with wearable recording kits, such that the streaming EEG signals are processed within a cloud environment.