Scholarlink

This paper presents an AI-based decision support system for spinal pathology diagnostics using MRI. The system incorporates an ensemble of neural networks and explainable AI (XAI) tools based on Grad-CAM. Our approach is aimed not only at enhancing the transparency of AI predictions, but also at improving clinical decisions in diagnostically complex cases. We experimentally show that (1) XAI can be used to restructure the training dataset to improve model performance, and (2) radiologists make more accurate diagnoses when provided with XAI maps alongside standard images. Our system shows promising results in detecting borderline cases of intervertebral disc protrusions, and lays the foundation for integrating XAI into clinical practice.

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

Explainable Artificial Intelligence for Doctors Decision Support in Diagnosing Spinal Pathologies

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Aleksandra Vatian,
  • Alexey Zubanenko,
  • Pavel Ulyanov,
  • Alexander Golubev,
  • Artem Beresnev,
  • Natalia Gusarova

摘要

This paper presents an AI-based decision support system for spinal pathology diagnostics using MRI. The system incorporates an ensemble of neural networks and explainable AI (XAI) tools based on Grad-CAM. Our approach is aimed not only at enhancing the transparency of AI predictions, but also at improving clinical decisions in diagnostically complex cases. We experimentally show that (1) XAI can be used to restructure the training dataset to improve model performance, and (2) radiologists make more accurate diagnoses when provided with XAI maps alongside standard images. Our system shows promising results in detecting borderline cases of intervertebral disc protrusions, and lays the foundation for integrating XAI into clinical practice.