Stroke Locus Net: Occluded Vessel Localization from MRI Modalities

2025-10-11

AI Summary

Precise localization of occluded vessels in ischemic stroke MRI remains challenging, as existing methods primarily focus on lesion segmentation and lack end-to-end vessel localization capability. This paper proposes the first end-to-end deep learning framework relying solely on T1-weighted MRI—integrating nnUNet-based lesion segmentation, atlas-guided arterial mapping, and conditional generative adversarial network (cGAN)-driven MRA synthesis to enable occlusion localization without acquired MRA. Crucially, it unifies lesion representation, anatomical priors, and cross-modal generation within a single model, overcoming the conventional reliance on multi-modal acquisitions. Evaluated on clinical T1 MRI data, the method achieves significantly higher localization accuracy than baseline approaches. It delivers an interpretable, low-barrier AI solution for rapid stroke diagnosis and treatment decision support.

Technology Category

• Computer Vision: Segmentation
• Intelligent Robots: Localization, Mapping, and Navigation
• Machine Learning: Neuro-Symbolic Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

A key challenge in ischemic stroke diagnosis using medical imaging is the accurate localization of the occluded vessel. Current machine learning methods in focus primarily on lesion segmentation, with limited work on vessel localization. In this study, we introduce Stroke Locus Net, an end-to-end deep learning pipeline for detection, segmentation, and occluded vessel localization using only MRI scans. The proposed system combines a segmentation branch using nnUNet for lesion detection with an arterial atlas for vessel mapping and identification, and a generation branch using pGAN to synthesize MRA images from MRI. Our implementation demonstrates promising results in localizing occluded vessels on stroke-affected T1 MRI scans, with potential for faster and more informed stroke diagnosis.

Problem

• Accurately localizing occluded vessels in ischemic stroke diagnosis
• Addressing limited machine learning methods for vessel localization
• Detecting and segmenting occluded vessels using only MRI scans

Innovation

• Combines nnUNet segmentation with arterial atlas mapping
• Uses pGAN to synthesize MRA images from MRI
• Provides end-to-end occluded vessel localization pipeline