Automatic classification of brain magnetic resonance images with hypercolumn deep features and machine learning

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Early detection of brain tumors is critical for patient outcomes.
• Magnetic resonance imaging (MRI) is a key diagnostic tool.
• Expert decision support systems can improve diagnostic accuracy in clinical settings.

Purpose of the Study:

• To develop and evaluate a computer-aided system for brain tumor detection.
• To leverage deep learning features for enhanced diagnostic performance.
• To compare the efficacy of Random Forest and Logistic Regression classifiers.

Main Methods:

• Utilized VGG16 convolutional layers to extract hypercolumn deep features.
• Identified keypoints within brain MRI scans.
• Employed Random Forest and Logistic Regression classifiers for diagnosis.
• Performed fivefold cross-validation for performance assessment.

Main Results:

• The Random Forest classifier achieved 94.51% accuracy.
• Sensitivity was 91.61%, with a false-negative rate of 8.39%.
• Specificity reached 97.42%, and precision was 97.29%.

Conclusions:

• The proposed model demonstrates high performance in brain tumor detection.
• Integration into computer-aided diagnosis systems could support clinical experts.
• This approach offers a promising tool for enhancing brain MRI analysis.