Deep learning classifies cell death in cerebral ischemia reperfusion injury

Le résumé fourni par la source

Cerebral ischemia–reperfusion injury (CIRI) is characterized by complex and overlapping cell death processes, including autophagy, apoptosis, and necrosis. Accurate discrimination among these cell death modalities is essential for pathological investigation, yet conventional pathological methods remain time-consuming and limited in precision. The present study aimed to develop a deep learning-based approach for the automated identification of autophagy, apoptosis, and necrosis following CIRI, thereby improving the efficiency and accuracy of pathological classification. A convolutional neural network (CNN)-based deep learning model was established to classify major cell death modalities after CIRI. On the independent test set, the model achieved an accuracy exceeding 93%, an average F1 score greater than 0.92, and a macro-averaged area under the ROC curve (mAUC) above 0.95. Gradient-weighted Class Activation Mapping (Grad-CAM) visualization revealed that the model focused on biologically relevant structural regions associated with distinct cell death patterns. Additional validation using multiple independent datasets further confirmed the robustness, reliability, and generalization capability of the proposed model. We successfully developed a deep learning-based system that automatically and accurately discriminates between autophagy, apoptosis, and necrosis in CIRI. The proposed approach offers an efficient and precise analytical tool for pathological research on CIRI and may facilitate mechanistic studies, accelerate therapeutic target identification, and provide theoretical support for future clinical intervention strategies.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.