Scholarlink

<p>Determining the specific type of brain tumor in a patient may be challenging due to the variability in their shapes and sizes. The utilization of MRI data, convolutional neural networks (CNNs), transformer-based architectures, and hybrid models facilitates the automatic classification and segmentation of brain tumors. This systematic literature analysis examines research on deep learning and explainable AI (xAI) from 2019 to 2025. PRISMA analyzed, categorized, and contrasted 90 peer-reviewed articles according to their architecture, performance metrics, and clinical significance. The evaluation highlights the shift from artisanal techniques to advanced transformer and xAI-augmented models that improve accuracy and clarity. Views on data diversity, multimodal learning, and practicable explainability frameworks address critical issues including generalizability, real-time implementation, and model transparency.</p>

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

A systematic literature review on advances in brain tumor detection using deep learning and explainable AI methods

  • 期刊论文

  • 发表日期：

    2025年11月20日

查看全文
  • Sara Tehsin,
  • Inzamam Mashood Nasir,
  • Robertas Damaševičius

摘要

Determining the specific type of brain tumor in a patient may be challenging due to the variability in their shapes and sizes. The utilization of MRI data, convolutional neural networks (CNNs), transformer-based architectures, and hybrid models facilitates the automatic classification and segmentation of brain tumors. This systematic literature analysis examines research on deep learning and explainable AI (xAI) from 2019 to 2025. PRISMA analyzed, categorized, and contrasted 90 peer-reviewed articles according to their architecture, performance metrics, and clinical significance. The evaluation highlights the shift from artisanal techniques to advanced transformer and xAI-augmented models that improve accuracy and clarity. Views on data diversity, multimodal learning, and practicable explainability frameworks address critical issues including generalizability, real-time implementation, and model transparency.