Refined Automatic Brain Tumor Classification Using Hybrid Convolutional Neural Networks for MRI Scans.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Accurate brain tumor classification from MRI scans is crucial for effective treatment planning.
• Deep learning models show promise in medical image analysis but require refinement for optimal performance.

Purpose of the Study:

• To develop and evaluate refined hybrid convolutional neural networks for classifying brain tumors (gliomas, meningiomas, pituitary tumors, and no tumor) from MRI scans.
• To compare the performance of fine-tuned pre-trained networks with novel hybrid architectures.

Main Methods:

• Utilized a dataset of 2880 T1-weighted contrast-enhanced MRI brain scans.
• Employed fine-tuned GoogleNet and AlexNet models for initial classification.
• Developed and tested hybrid networks: AlexNet-SVM and AlexNet-KNN, integrating Support Vector Machine and K-Nearest Neighbors classifiers with AlexNet.

Main Results:

• Fine-tuned GoogleNet and AlexNet achieved validation accuracies of 91.5% and 90.21%, respectively.
• Hybrid AlexNet-SVM and AlexNet-KNN models demonstrated improved performance, reaching 96.9% and 98.6% validation accuracy.
• Testing on a separate dataset confirmed AlexNet-KNN's superior performance with 97% accuracy.

Conclusions:

• The proposed AlexNet-KNN hybrid network offers a highly accurate method for automatic brain tumor classification from MRI scans.
• This approach has the potential to significantly reduce clinical diagnosis time and aid in early detection.