A deep ensemble learning framework for brain tumor classification using data balancing and fine-tuning

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Accurate brain tumor diagnosis is crucial for patient outcomes, but manual MRI analysis is time-consuming.
• Deep learning (DL) offers potential for efficient and accurate diagnostic assistance.
• Misclassification of brain tumors can lead to reduced life expectancy, highlighting the need for precise methods.

Purpose of the Study:

• To develop and evaluate an innovative deep ensemble approach for brain tumor classification using transfer learning (TL).
• To enhance diagnostic accuracy and efficiency in analyzing brain tumor MRI datasets.
• To compare the performance of the proposed model against existing state-of-the-art methods.

Main Methods:

• A deep ensemble model was developed using transfer learning (TL) architectures.
• Preprocessing, synthetic data generation (SDG), and fine-tuning of TL models were performed.
• Model weights were optimized using Genetic Algorithm-based Weight Optimization (GAWO) and Grid Search-based Weight Optimization (GSWO).

Main Results:

• The proposed deep ensemble model achieved high classification accuracies, with GSWO reaching 99.84%.
• Individual TL models like Xception and ResNet variants also showed strong performance (99.57% - 99.33%).
• The model demonstrated superiority over State of Arts (SOA) works in comparative analysis.

Conclusions:

• The optimized deep ensemble model is a robust and reliable tool for brain tumor classification.
• This approach can significantly assist neurologists and clinicians in making precise and timely diagnostic decisions.
• The study underscores the potential of advanced DL techniques in medical diagnostics.