CNN-ViT 交叉注意力融合模型用于急性脑卒中分类：多中心外部验证研究
Acute Stroke Classification Using Asymmetric Data Standardization and Orthogonal-Focal Loss Optimized CNN-ViT Cross-Attention Fusion: A Multicenter External Validation Study
AI 导读
一项研究提出结合 CNN 与 ViT 主干、由正交焦点损失优化的交叉注意力融合分级辅助诊断系统，用于快速区分缺血性与出血性脑卒中，内部数据集 F1 分数达 95.48%。
正文 · 原文
Abstract:Background/Objectives: Rapid differentiation of ischemic and hemorrhagic stroke subtypes is vital for acute management. Although current deep learning models achieve high single-dataset accuracy, they suffer from "shortcut learning" and domain shift in external validation. This study proposes a hierarchical computer-aided diagnostic (CAD) system mimicking emergency triage. Methods: We introduce a 3-channel Radiology-Focused Preprocessing pipeline that feeds radiological Hounsfield Unit (HU) windows into Convolutional Neural Network (CNN) and Vision Transformer (ViT) backbones via asymmetric formats (.mat and .png). The architecture utilizes a novel joint cross-attention fusion network optimized by an Integrated Orthogonal-Focal Loss. Results: Our system achieved a 95.48% F1-score on the internal stroke dataset. However, multi-center external validation (AISD, BHSD, PhysioNet) using a Mask Anchor filter revealed a drop to the 58-62% range, exposing the "internal validation illusion" caused by hardware noise. Critically, models exhibited high precision (77%) but low sensitivity (57%) for ischemic stroke, proving prone to hemorrhage overdiagnosis. Multi-seed variance analysis demonstrated that the DenseNet201+ViT backbone yielded the lowest deviation (0.59) and highest stability, making it the most reliable architecture for real-world clinical integration. Conclusions: To overcome diagnostic challenges, future research must focus on multimodal architectures that combine 3D volumetric extraction with clinical data.
| Comments: | All research data has been uploaded to Zenodo (DOI: https://doi.org/10.5281/zenodo.22132268). The data presented in this study are available on request from the corresponding author due to ongoing research studies |
| Subjects: | Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.34529 [eess.IV] |
| (or arXiv:2610.34529v1 [eess.IV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.34529 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tansel Uyar Dr. [view email]
[v1]
Sun, 25 Oct 2026 19:43:28 UTC (1,202 KB)
来源：arXiv cs.AI · arxiv.org