Scholarlink

The work highly enhances the ability to segment brain tumors in 2D MRI scans using DeepLabV3+ model with ResNet-152V2 backbone, that provides high Dice coefficient scores of precision. In order to enhance transparency and develop trust, we use Explainable AI methods such as Grad-CAM, Grad-CAM++, XGrad-CAM, and LRP. These techniques offer excellent visualizations of the areas affecting the decision made by the model, which is highly beneficial in getting insights on the way the model works. The results obtained by us indicate the usefulness of the combination of DeepLabV3 + ResNet-152V2 with Explainable AI to make the brain tumor segmentation approach reliable and understandable. Such practice increases the precision of diagnostic work and facilitates successful treatment practices in medical conditions, and that will lead to the improved prognosis of a patient and the development of medical procedures regarding the use of images in the medical process.

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

Leveraging DeepLabV3+ with Explainable AI for Brain Tumor Segmentation in 2D MRI

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jiten Ganwani,
  • Samruddhi M. Kolekar,
  • Dhananjay R. Kalbande,
  • Maheshkumar H. Kolekar

摘要

The work highly enhances the ability to segment brain tumors in 2D MRI scans using DeepLabV3+ model with ResNet-152V2 backbone, that provides high Dice coefficient scores of precision. In order to enhance transparency and develop trust, we use Explainable AI methods such as Grad-CAM, Grad-CAM++, XGrad-CAM, and LRP. These techniques offer excellent visualizations of the areas affecting the decision made by the model, which is highly beneficial in getting insights on the way the model works. The results obtained by us indicate the usefulness of the combination of DeepLabV3 + ResNet-152V2 with Explainable AI to make the brain tumor segmentation approach reliable and understandable. Such practice increases the precision of diagnostic work and facilitates successful treatment practices in medical conditions, and that will lead to the improved prognosis of a patient and the development of medical procedures regarding the use of images in the medical process.