AsymFeX: A Symmetry-Driven Framework for Ischemic Stroke Segmentation Across Imaging Modalities and Stroke Stages

Maunil Shah; Vaanathi Sundaresan

eess.IVAug 20, 2026

AsymFeX: A Symmetry-Driven Framework for Ischemic Stroke Segmentation Across Imaging Modalities and Stroke Stages

Authors: Maunil Shah, Vaanathi Sundaresan

Organizations: Department of Computational and Data Sciences, Indian Institute of Science, Bengaluru 560012, India

Abstract

Fast and accurate segmentation of Acute Ischemic Stroke (AIS) lesions is essential for stroke prognosis and treatment planning. Non-contrast CT (NCCT), the first-line imaging modality for diagnosing ischemic infarcts, exhibits subtle infarct contrast, making manual delineation slow and labor-intensive. Motivated by this, and by the clinical practice of comparing brain hemispheres to localize infarcts, we propose a two-stage, nnU-Net-compatible 3D segmentation method. The first stage corrects head tilt to align each scan to its true anatomical mid-sagittal plane; the second applies a novel Asymmetric Feature Extraction (AsymFeX) module, comparing each voxel to its true contralateral counterpart within a local 3 x 3 x 3 neighborhood via cross-hemispheric attention, feature disparity estimation, and dual-scale gating to capture both large and small infarcts. On AISD, our method achieves 0.6796 Dice, 23.53 mm HD95, and 7.69 mL AVD, significantly outperforming existing state-of-the-art methods, with clinically relevant volumetric analysis at the 70 mL thrombolysis-eligibility threshold. Proof-of-concept evaluation on ATLAS v2.1 and ISLES'24 demonstrates that the same symmetry-driven design generalizes across imaging modalities and stroke time points without architectural changes, further supported by an uncertainty analysis assessing reliability under clinical deployment. Code is publicly available at https://github.com/biomedia-lab/AIS-detection.

Figures & tables

Figure 1: Illustration of Stage 2 of the proposed methodology: Stage 2 receives $V_{final}$ and $V_{flipped}$ from stage 1 as the input and uses AsymFex blocks along the deeper layers for leveraging the asymmetric differences. The AsymFex comprises of 3 sub-modules, viz., Cross-Stream Alignment (CSA), Feature Disparity Estimation (FDE) and Dual-Scale Gating (DSG).

Figure 2: Qualitative comparison of segmentation outputs: Visual results of comparison of our method with existing methods along with Uncertainty maps on ATLASv2.1 (row 1), ISLES’24 (row 2 - Penumbra, row 3 - Core) and AISD (row 4).

Figure 3: Attention cost versus input size ( $N=D\!\times\!H\!\times\!W$ ) for full attention (red) and CSA (blue). Full attention scales quadratically ( $\mathcal{O}(N^{2})$ ) while our CSA scales linearly ( $\mathcal{O}(N)$ ).

Metric	Local (Ours)	Global	Ratio
Parameters	263,168	263,168	1.0 $\times$
Total FLOPs (G)	7.93	226.19	28.5 $\times$
Attention FLOPs (G)	0.41	218.67	531 $\times$
Inference time (ms)	6.17	72.42	11.7 $\times$
Training time (ms)	27.39	227.27	8.3 $\times$
Inference peak memory (GB)	1.25	12.36	9.9 $\times$

Table 2: Cost of local (ours) vs. full cross-attention at

14\times 32\times 32

C=256

(

N=14{,}336

), batch size 1, single GPU. Ratio = global / local. FLOPs use MAC

=2

operations.

Figure 4: Volumetric Analysis for AISD (top) Bland-Altman plot comparing predicted volume with the ground truth, (bottom) Scatter plot of the predicted volumes versus the ground truth.

Figure 5: Calibration plots. Bars show the observed foreground frequency per bin, red markers show the mean predicted probability and diagonal denotes perfect calibration. Deviations from the bars indicate miscalibration. The corresponding ECE values are reported in Table 5 .

Table 4: Quantitative comparison of the proposed method against existing approaches on ATLAS v2.1 and ISLES’24. Symbol * denotes statistically significant difference in performance with p-value

<

0.05 using Wilcoxon signed rank test.

Metric	AISD	ATLAS v2.1	ISLES’24
Metric	AISD	ATLAS v2.1	Infarct	Penumbra
AUROC	0.8102	0.7277	0.7706	0.7053
AUPRC	0.3678	0.3832	0.4819	0.4357
Prevalence	0.1296	0.1947	0.2515	0.2837
Mean uncertainty
TP	0.1719	0.2206	0.3028	0.4545

Table 5: Uncertainty metrics per dataset. All values are computed within the foreground ROI. Every metric is calculated over the MC-averaged T=30 passes probabilities.

Explore similar work

Ischemic Stroke Segmentation and Net Water Uptake Quantification on Multicenter Non-Contrast CT Using Supervised Target-Domain Adaptation
Sep 17, 2026Linus Britt, Maximilian Nielsen, Susan Klapproth +5Medical Image Segmentation
Institute for Applied Medical Informatics, University Medical Center Hamburg-Eppendorf, Hamburg, Germany · Institute of Computational Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Germany · Department of Diagnostic and Interventional Neuroradiology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany +4
Clinically-aligned ischemic stroke segmentation and ASPECTS scoring on NCCT imaging using a slice-gated loss on foundation representations
Feb 27, 2026Hiba Azeem, Behraj Khan, Tahir Qasim SyedMedical Image Segmentation Computed Tomography
Institute of Business Administration Karachi
Deep Learning Segmentation of Diffusion-Weighted MRI Acute Ischaemic Stroke: A Pragmatic Evaluation Across Three Datasets
Aug 26, 2026Atle Bjørnerud, Till Schellhorn, Thor H. Skattør +4Medical Image Segmentation U-Net
Computational Radiology & Artificial Intelligence Unit, Department of Physics and Computational Radiology, Oslo University Hospital, Oslo, Norway · Center for Lifespan Changes in Brain and Cognition, Department of Psychology, University of Oslo, Oslo, Norway · Division of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo, Norway +5

Sep 17, 2026J/K move · Enter open · S save

Ischemic Stroke Segmentation and Net Water Uptake Quantification on Multicenter Non-Contrast CT Using Supervised Target-Domain Adaptation

Linus Britt, Maximilian Nielsen, Susan Klapproth, Andre Kemmling+4

Institute for Applied Medical Informatics, University Medical Center Hamburg-Eppendorf, Hamburg, Germany · Institute of Computational Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Germany · Department of Diagnostic and Interventional Neuroradiology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany+4

Medical Image Segmentation

Open paper AskPDF arXiv

Objectives: Quantitative assessment of infarct hypodensity on non-contrast computed tomography (NCCT), including net water uptake (NWU), requires manual or semi-manual lesion delineation, often guided by CT perfusion or diffusion-weighted MRI, limiting clinical applicability. Automated segmentation on NCCT could enable efficient biomarker extraction such as NWU but remains challenging across heterogeneous multicenter data. This study aimed to develop and externally test a domain-aware deep learning framework for ischemic stroke segmentation on NCCT and assess its suitability for NWU quantification. Materials & Methods: In this retrospective multicenter study of 801 patients from four datasets, an nnU-Net-based model was trained on NCCT scans from the University Medical Center Hamburg-Eppendorf and the Acute Ischemic Stroke Dataset. To adapt to new domains, the model was fine-tuned on target-domain subsets from Boston (n=11) and ISLES (n=75), with evaluation on held-out cases not used for fine-tuning. Automated segmentations and NWU values were compared with expert references. Results: For lesions

\geq

30 mL, median Dice was 0.68 (Boston) and 0.56 (ISLES). Including smaller lesions, which predominated in ISLES, median Dice was 0.54 (interquartile range [IQR] 0.30-0.70) for acute lesion segmentation (Boston dataset) and 0.20 (IQR 0.03-0.41) for NCCT lesion segmentations when compared to post-treatment infarct (primary target of the ISLES challenge). Automated NWU mean absolute error was 1.37 percentage points (SD 1.61, Boston). Conclusion: Target-domain adaptation supported NCCT-only infarct segmentation across heterogeneous external cohorts, although performance varied across domains. The approach enabled low-error NWU quantification from baseline NCCT without advanced imaging, supporting further prospective clinical evaluation.

Abstract

Explore similar work

Ischemic Stroke Segmentation and Net Water Uptake Quantification on Multicenter Non-Contrast CT Using Supervised Target-Domain Adaptation

Clinically-aligned ischemic stroke segmentation and ASPECTS scoring on NCCT imaging using a slice-gated loss on foundation representations

Deep Learning Segmentation of Diffusion-Weighted MRI Acute Ischaemic Stroke: A Pragmatic Evaluation Across Three Datasets