Scholarlink

Medical image segmentation plays a critical role in accurately localizing tumor regions within MRI scans. With the advancement of deep learning techniques, state-of-the-art models have achieved high accuracy in delineating tumor boundaries. However, in the context of medical imaging, it is imperative not only to identify pathological regions but also to provide interpretable justifications for such predictions. In this work, we propose a suite of explainability-driven methodologies tailored for medical image segmentation. Furthermore, we introduce a quantitative approach to evaluate the alignment between model explanations and expert annotations using the Intersection over Union (IoU) metric. Experimental evaluations on standard benchmarks demonstrate the effectiveness of the proposed framework in achieving both accurate and interpretable segmentation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Explainability-Driven Multimodal Glioma Segmentation Using a Hybrid CNN-Transformer Architecture and IoU-Based Attribution Analysis

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Gokul Singh Chauhan,
  • Meghna Kapoor,
  • Badri Narayan Subudhi

摘要

Medical image segmentation plays a critical role in accurately localizing tumor regions within MRI scans. With the advancement of deep learning techniques, state-of-the-art models have achieved high accuracy in delineating tumor boundaries. However, in the context of medical imaging, it is imperative not only to identify pathological regions but also to provide interpretable justifications for such predictions. In this work, we propose a suite of explainability-driven methodologies tailored for medical image segmentation. Furthermore, we introduce a quantitative approach to evaluate the alignment between model explanations and expert annotations using the Intersection over Union (IoU) metric. Experimental evaluations on standard benchmarks demonstrate the effectiveness of the proposed framework in achieving both accurate and interpretable segmentation.