A Novel and Effective Brain Tumor Classification Model Using Deep Feature Fusion and Famous Machine Learning...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Brain tumors are a significant global health concern, with manual image analysis for diagnosis being time-consuming and error-prone.
• Existing automated brain tumor detection methods often suffer from low accuracy and high false-positive rates.
• Accurate and efficient brain tumor identification and classification are crucial for timely treatment and improved patient outcomes.

Purpose of the Study:

• To develop a novel, highly accurate multiclass brain tumor classification method using deep feature fusion.
• To enhance the robustness and reliability of automated brain tumor detection from Magnetic Resonance Images (MRIs).
• To improve upon the performance of existing brain tumor classification systems.

Main Methods:

• Preprocessed MR images using min-max normalization and applied extensive data augmentation.
• Extracted deep convolutional neural network (CNN) features from transfer learning architectures (AlexNet, GoogLeNet, ResNet18).
• Fused extracted deep features into a single vector, then classified using Support Vector Machine (SVM) and K-nearest neighbor (KNN).

Main Results:

• The fused feature vector significantly outperformed individual feature vectors in classification performance.
• The proposed deep feature fusion method achieved a high accuracy of 99.7% on a dataset of 15,320 MRIs.
• The novel technique demonstrated superior performance compared to existing brain tumor classification systems.

Conclusions:

• Deep feature fusion is an effective strategy for improving the accuracy of multiclass brain tumor classification from MRIs.
• The proposed method offers a promising, highly accurate, and efficient tool for clinical application in brain tumor diagnosis.
• This advanced AI approach can aid medical professionals in faster and more reliable identification of brain tumors.