Interpretable Deep Learning for Stroke Diagnosis: A Framework for Classification and Visualization on Brain CT Scans
Keywords:
troke Classification, Explainable AI (XAI), Deep Learning, Computed Tomography, Medical Imaging, Decision Support Systems, Grad-CAMAbstract
Background: Rapid and accurate diagnosis of stroke from non-contrast computed tomography (CT) scans is paramount for effective treatment, yet it faces challenges including diagnostic subtlety and inter-observer variability. While deep learning models offer promising solutions, their "black box" nature often impedes clinical adoption due to a lack of transparency and trust. This study addresses this gap by proposing and validating a comprehensive Explainable AI (XAI) framework for stroke classification.
Methods: We developed an integrated framework comprising a Convolutional Neural Network (CNN) for classifying brain CT scans into three categories: ischemic stroke, hemorrhagic stroke, and normal. The framework's classification model was trained and validated on a dataset of thousands of CT images, sourced from the TEKNOFEST-2021 Stroke Data Set [17]. To ensure model transparency, we integrated the Grad-CAM method to generate visual saliency maps that highlight the image regions most influential in the model's decision-making process. Model performance was evaluated using accuracy, precision, recall, F1-score, and the Area Under the Curve (AUC).
Results: The proposed classification model achieved high diagnostic performance, with an overall accuracy of 96.2% and an AUC of 0.989. The qualitative analysis demonstrated that the XAI module successfully produced clinically coherent saliency maps. For hemorrhagic and ischemic cases, the generated heatmaps accurately localized the pathological areas, aligning with radiological findings and providing a clear basis for the model's predictions.
Conclusion: Our findings suggest that the proposed XAI framework can function as a reliable and transparent decision-support tool. By combining high classification accuracy with intuitive visual explanations, it has the potential to enhance diagnostic confidence, assist clinicians in acute settings, and foster greater trust in the application of artificial intelligence in stroke care.
References
Aksoy, S.; Demircioglu, P.; Bogrekci, I. Optimizing Stroke Classification with Pre-Trained Deep Learning Models. J. Vasc. Dis. 2024, 3, 480–494.
Salman, S.; Gu, Q.; Dherin, B.; Reddy, S.; Vanderboom, P.; Sharma, R.; Lancaster, L.; Tawk, R.; Freeman, W.D. Hemorrhage Evaluation and Detector System for Underserved Populations: HEADS-UP. Mayo Clin. Proc. Digit. Health 2023, 1, 547–556.
Hossain, M.; Ahmed, M.; Nafi, A.A.N.; Islam, R.; Ali, S.; Haque, J.; Miah, S.; Rahman, M.; Islam, K. A Novel Hybrid ViT-LSTM Model with Explainable AI for Brain Stroke Detection and Classification in CT Images: A Case Study of Rajshahi Region. Comput. Biol. Med. 2025, 186, 109711.
Abdi, H.; Sattar, M.U.; Hasan, R.; Dattana, V.; Mahmood, S. Stroke Detection in Brain CT Images Using Convolutional Neural Networks: Model Development, Optimization and Interpretability. Information 2025, 16, 345.
Patil, S.; Rossi, R.; Jabrah, D.; Doyle, K. Detection, Diagnosis and Treatment of Acute Ischemic Stroke: Current and Future Perspectives. Front. Med. Technol. 2022, 4, 748949.
Wang, M.; Lin, Y.; Gu, F.; Xing, W.; Li, B.; Jiang, X.; Liu, C.; Li, D.; Li, Y.; Wu, Y.; et al. Diagnosis of Cognitive and Motor Disorders Levels in Stroke Patients through Explainable Machine Learning Based on MRI. Med. Phys. 2024, 51, 1763–1774.
Patel, C.H.; Undaviya, D.; Dave, H.; Degadwala, S.; Vyas, D. EfficientNetB0 for Brain Stroke Classification on Computed Tomography Scan. In Proceedings of the 2023 2nd International Conference on Applied Artificial Intelligence and Computing (ICAAIC), Salem, India, 4 May 2023.
Foroushani, H.M.; Hamzehloo, A.; Kumar, A.; Chen, Y.; Heitsch, L.; Slowik, A.; Strbian, D.; Lee, J.-M.; Marcus, D.S.; Dhar, R. Accelerating Prediction of Malignant Cerebral Edema After Ischemic Stroke with Automated Image Analysis and Explainable Neural Networks. Neurocrit Care 2022,1 36, 471–482.
Jabal, M.S.; Joly, O.; Kallmes, D.; Harston, G.; Rabinstein, A.; Huynh, T.; Brinjikji, W. Interpretable Machine Learning Modeling for Ischemic Stroke Outcome Prediction. Front. Neurol. 2022, 13, 884693.
Connie, T.; Tan, Y.F.; Goh, M.K.O.; Hon, H.W.; Kadim, Z.; Wong, L.P. Explainable Health Prediction from Facial Features with Transfer Learning. IFS 2022, 42, 2491–2503.
Brima, Y.; Atemkeng, M. Saliency-Driven Explainable Deep Learning in Medical Imaging: Bridging Visual Explainability and Statistical Quantitative Analysis. BioData Min. 2024, 17, 18.
Kim, J.G.; Ha, S.Y.; Kang, Y.-R.; Hong, H.; Kim, D.; Lee, M.; Sunwoo, L.; Ryu, W.-S.; Kim, J.-T. Automated Detection of Large Vessel Occlusion Using Deep Learning: A Pivotal Multicenter Study and Reader Performance Study. J. NeuroIntervent Surg. 2024, jnis-2024-022254.
Heo, J. Application of Artificial Intelligence in Acute Ischemic Stroke: A Scoping Review. Neurointervention 2025, 20, 4–14.
Riedel, E.O.; de la Rosa, E.; Baran, T.A.; Petzsche, M.H.; Baazaoui, H.; Yang, K.; Musio, F.A.; Huang, H.; Robben, D.; Seia, J.O.; et al. ISLES’24—A Real-World Longitudinal Multimodal Stroke Dataset. arXiv 2024, arXiv:2408.11142.
Al-Janabi, O.M.; El Refaei, A.; Elgazzar, T.; Mahmood, Y.M.; Bakir, D.; Gajjar, A.; Alateya, A.; Jha, S.K.; Ghozy, S.; Kallmes, D.F.; et al. Current Stroke Solutions Using Artificial Intelligence: A Review of the Literature. Brain Sci. 2024, 14, 1182.
Heo, J.; Ryu, W.-S.; Chung, J.-W.; Kim, C.K.; Kim, J.-T.; Lee, M.; Kim, D.; Sunwoo, L.; Ospel, J.M.; Singh, N.; et al. Automated Ischemic Stroke Lesion Detection on Non-Contrast Brain CT: A Large-Scale Clinical Feasibility Test. 2025.
Koc, U.; Akcapinar Sezer, E.; Ozkaya, Y.A.; Yarbay, Y.; Taydas, O.; Ayyildiz, V.A.; Kiziloglu, H.A.; Kesimal, U.; Cankaya, I.; Besler, M.S.; et al. Artificial Intelligence in Healthcare Competition (TEKNOFEST-2021): Stroke Data Set. Eurasian J. Med. 2022, 54, 248–258.
Downloads
Published
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors retain the copyright of their articles published in this journal. All articles are licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0). This license permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly cited.