A Hybrid Approach for Detection and Classification of Brain Stroke Using Non-Contrast CT Imaging and EEG Time Series Data

View at publisher Summarise with VIRAT
Authors: Rahul Chiranjeevi. V, G Priyadharshini, Pon Madhumitha A, Nikitha R
Cited by: 0
References: 16

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

In recent years brain stroke is identified as a major neurological condition among 60 million people worldwide due to their abnormal brain activity patterns. Early detection of stroke is critical in determining accurate treatment and reducing morbidity. The Proposed research utilizes a hybrid deep learning detection model by combining Three-dimensional Residual Network (3D-ResNet 18) and Graph Convolutional Network (GCN) to perform classification and Detection. A unified cross-modal attention and gated-fusion mechanism is designed thereby improving the detection and classification accuracy The system uses Non-Contrast CT Images and EEG time series data to improve the detection accuracy. Various preprocessing techniques are employed on both CT images and EEG data to remove noise, null values, bad-channel interpolations, and normalization to train the model. Spatial and temporal feature vectors are obtained from the model and fused using gated fusion mechanism for further classification. Cross Modal attention is performed to create context-aware interaction between CT and EEG data. The temporal signals are mapped with corresponding spatial values to detect the stroke. Experiments are performed on multimodal dataset containing both CT and EEG data. Results demonstrates that proposed model achieved a significant accuracy of 94.2 % with a precision of 93.1 %, Recall values of 92.2 % and$\mathbf{F 1}$-score of 0.94. Further lesion masks on brain images are generated for lesion identification. Proposed model demonstrates a significant improvement in detecting brain stroke in Real-time in comparison with state-of-art methods.

Fields of Study & Subjects

Acute Ischemic Stroke Management EEG and Brain-Computer Interfaces Artificial Intelligence in Healthcare

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1–7
Publication date
21 January 2026
Language
EN