Brain Tumor Classification Model Guided by Class Activation Mapping

doi:10.1016/j.compmedimag.2026.102726

Brain Tumor Classification Model Guided by Class Activation Mapping

2026-02-09

0

PRE

AI

DOI：10.1016/j.compmedimag.2026.102726

Original

Original request for help

Save

Abstract

En 中文

• We propose CAM-BTC, a brain tumor classification model that uses Class Activation Mapping for enhanced accuracy and interpretability. • The model incorporates an innovative two-branch structure, including an attention branch with a Saliency Learning Module and a perception branch with a Sample Selection Module, enhancing both the model's performance and interpretability. • The model achieves an impressive accuracy range of 96-99%, with an average of 97.41%, surpassing other deep learning methods for brain tumor classification.

Keywords:

CAM-BTC

Brain Tumor Classification

Class Activation Mapping

Attention Branch

Saliency Learning Module

Brain Tumor Classification Model Guided by Class Activation Mapping

Abstract

Journal

Organization