Related Experiment Video
Updated: May 20, 2025
09:47
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
931
Brain tumor intelligent diagnosis based on Auto-Encoder and U-Net feature extraction
Yaru Cao1, Fengning Liang1, Teng Zhao1
1School of Medical Information and Engineering, Xuzhou Medical University, Xuzhou, Jiangsu, China.
Plos One
|March 24, 2025
Summary
This study introduces an automated brain tumor classification system using improved U-Net and CRNN models. The AI approach enhances diagnostic accuracy for glioma grading, IDH1 mutation status, and pituitary tumors.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Accurate preoperative brain tumor classification is crucial for personalized treatment.
- Current manual methods face challenges in efficiency and accuracy, risking misdiagnosis.
Purpose of the Study:
- To develop a fully automated approach for brain tumor classification using magnetic resonance imaging (MRI).
- To improve diagnostic accuracy and efficiency compared to existing methods.
Main Methods:
- A novel approach combining an improved U-Net feature extractor with a convolutional recurrent neural network (CRNN) classifier.
- The U-Net encoder utilizes dense blocks for enhanced feature propagation, while the decoder uses residual blocks to prevent gradient disappearance.
- Skip connections merge low-level and high-level features for comprehensive analysis.
Main Results:
- The model achieved high accuracy in classifying glioma (90.72%), glioma IDH1 mutation status (94.35%), and pituitary tumor texture (94.64%).
- Performance was validated on local hospital data and TCIA glioma imaging data.
Conclusions:
- The proposed automated system demonstrates superior accuracy in brain tumor classification.
- This AI-driven approach holds significant potential for improving clinical diagnosis and treatment planning.
