Epileptic Seizure Detection in SEEG Signals Using a Signal Embedding Temporal-Spatial-Spectral Transformer Model

High-performance methods for automated detection of epileptic stereo-electroencephalography (SEEG) have important clinical research implications, improving the diagnostic efficiency and reducing physician burden. However, few studies have been able to consider the process of seizure propagation, thus failing to fully capture the deep representations and variations in SEEG in the temporal, spatial, and spectral domains. In this article, we construct a novel long-term SEEG seizure dataset (XJSZ dataset) and propose signal embedding temporal-spatial-spectral transformer (SE-TSS-Transformer) framework. First, we design signal embedding (SE) module to reduce feature dimensions and adaptively construct optimal representation for subsequent analysis. Second, we integrate unified multiscale temporal-spatial-spectral (TSS) analysis to capture multilevel, multidomain deep features. Finally, we use the transformer encoder to learn the global relevance of features, enhancing the network's ability to express SEEG features. Experimental results demonstrate state-of-the-art detection performance on the XJSZ dataset, achieving sensitivity, specificity, and accuracy of 99.03%, 99.34%, and 99.03%, respectively. Furthermore, we validate the scalability of the proposed framework on two public datasets of different signal sources, demonstrating the power of the SE-TSS-Transformer framework for capturing diverse multiscale TSS patterns in seizure detection.