StrokeNeXt: A Siamese-encoder Approach for Brain Stroke Classification in Computed Tomography Imagery

TL;DR

StrokeNeXt tackles CT-based brain stroke classification with a dual-branch ConvNeXt encoder setup whose embeddings are merged by a lightweight 1D fusion decoder. The approach achieves state-of-the-art accuracy and calibration for both stroke presence (non-stroke vs stroke) and subtype (ischemic vs hemorrhagic) tasks, with F1-scores approaching 0.99 and MCC around 0.97, across a real-world dataset of 6,774 images. Statistical tests (McNemar) confirm significant improvements over baselines, and analyses show balanced per-class sensitivity and specificity along with fast inference and practical training times. The results demonstrate a favorable trade-off between diagnostic performance and computational efficiency, supporting deployment across varied resource settings while maintaining robust clinical reliability.

Abstract

We present StrokeNeXt, a model for stroke classification in 2D Computed Tomography (CT) images. StrokeNeXt employs a dual-branch design with two ConvNeXt encoders, whose features are fused through a lightweight convolutional decoder based on stacked 1D operations, including a bottleneck projection and transformation layers, and a compact classification head. The model is evaluated on a curated dataset of 6,774 CT images, addressing both stroke detection and subtype classification between ischemic and hemorrhage cases. StrokeNeXt consistently outperforms convolutional and Transformer-based baselines, reaching accuracies and F1-scores of up to 0.988. Paired statistical tests confirm that the performance gains are statistically significant, while class-wise sensitivity and specificity demonstrate robust behavior across diagnostic categories. Calibration analysis shows reduced prediction error compared to competing methods, and confusion matrix results indicate low misclassification rates. In addition, the model exhibits low inference time and fast convergence.

Figures (6)

• Figure 1: Overview of the proposed StrokeNeXt architecture. The model processes the input CT image through two parallel ConvNeXt encoders. Their outputs are fused via a custom convolutional decoder composed of a stack operation, a 1D convolutional fusion path with transformation layers, and a classification head.
• Figure 2: Sample images from the used dataset in this work.
• Figure 3: ConvNeXt architecture (tiny version). Source: 10971457
• Figure 4: Structure of fusion decoder developed in this study.
• Figure 5: Confusion matrices comparison between StrokeNeXt and other models on stroke presence classification (non-stroke vs. stroke).