Brain Tumour Classification Model Based on Spatial Block-Residual Block Collaborative Architecture with Strip Pooling...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Accurate brain tumor classification is vital for effective diagnosis and treatment.
• Traditional methods are limited by the difficulty in obtaining precise tumor masks.

Purpose of the Study:

• To develop a brain tumor classification model that does not require tumor masks.
• To achieve high-precision, multi-scale feature representation for improved classification accuracy.

Main Methods:

• A novel cooperative architecture combining VGG spatial blocks and ResNet residual blocks.
• Striped pooling modules for effective multi-level feature fusion and cross-layer integration.
• End-to-end classification using a Softmax classifier on fused multi-scale features.

Main Results:

• The proposed model achieved 97.29% accuracy in brain tumor classification.
• Significantly outperformed traditional convolutional neural network methods.
• Demonstrated effectiveness in multi-scale feature learning without requiring tumor masks.

Conclusions:

• The developed model offers a promising approach for mask-free brain tumor classification.
• Its high accuracy and efficiency hold significant potential for clinical applications.
• Facilitates improved early diagnosis and treatment planning for brain tumors.