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

Hiba Azeem; Behraj Khan; Tahir Qasim Syed

eess.IVFeb 27, 2026

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

Authors: Hiba Azeem, Behraj Khan, Tahir Qasim Syed

Organizations: Institute of Business Administration Karachi

Abstract

Rapid infarct assessment on non-contrast CT (NCCT) is essential for acute ischemic stroke management. Most deep learning methods perform pixel-wise segmentation without modeling the structured anatomical reasoning underlying ASPECTS scoring, where basal ganglia (BG) and supraganglionic (SG) levels are clinically interpreted in a coupled manner. We propose a clinically aligned framework that combines a frozen DINOv3 backbone with a lightweight decoder and introduce a Territory-Aware Gated Loss (TAGL) to enforce BG-SG consistency during training. This anatomically informed supervision adds no inference-time complexity. Our method achieves a Dice score of 0.6385 on AISD, outperforming prior CNN and foundation-model baselines. On a proprietary ASPECTS dataset, TAGL improves mean Dice from 0.698 to 0.767. These results demonstrate that integrating foundation representations with structured clinical priors improves NCCT stroke segmentation and ASPECTS delineation.

Figures & tables

Panel of Figure 1 — Figure 1: Overview of the proposed framework. An NCCT slice is processed by a frozen DINOv3 backbone and a multi-scale feature extractor in parallel. Two DPT-style dual-arm decoders process the respective features, and their outputs are fused via a Territory-Aware Gated Loss (TAGL) module before producing the final multi-class segmentation map $\hat{Y}$ .

Figure 2: Training and validation performance curves. Left: Combined BCE–Dice loss. Right: Corresponding Dice score evolution. Model selection is based on the highest validation Dice.

Figure 3: Qualitative segmentation results on the AISD dataset produced by our best-performing model. Examples illustrate typical lesion appearances across challenging NCCT slices, including low-contrast regions and small or fragmented infarcts.

Model	Encoder Type	Dice	IoU
SEAN [ 11 ]	CNN	0.5784	–
UNETR [ 7 ]	Swin Transformer	0.4470	0.3510
SwinUNETR [ 6 ]	Swin Transformer	0.4470	0.3510
UNet++ [ 21 ]	CNN	0.4410	0.3520
DINOv3 benchmarking [ 20 ]	Frozen DINOv3	0.3674	0.2703
StrDiSeg (frozen) [ 3 ]	Frozen DINOv3	0.3720	0.2690

Table 1: Performance comparison on AISD. Best results in bold.

Configuration	Mean Dice	IoU
Baseline (CE only)	0.698	0.683
Proposed (CE + TAGL)	0.767	0.756

Table 2: Performance comparison on the proprietary ASPECTS dataset. Best results in bold.

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
AsymFeX: A Symmetry-Driven Framework for Ischemic Stroke Segmentation Across Imaging Modalities and Stroke Stages
Aug 20, 2026Maunil Shah, Vaanathi SundaresanMedical Image Segmentation 3D Medical Image Segmentation
Department of Computational and Data Sciences, Indian Institute of Science, Bengaluru 560012, India
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

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

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