Brain tumor classification in MRI image using convolutional neural network.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Brain tumors represent a critical health challenge, necessitating advanced diagnostic tools.
• Deep learning, particularly Convolutional Neural Networks (CNNs), has shown significant promise in medical image analysis and diagnostics.
• Accurate and efficient classification of brain MRI scans is crucial for timely medical diagnosis.

Purpose of the Study:

• To develop and evaluate a novel CNN model for classifying brain MRI scans as cancerous or non-cancerous.
• To compare the performance of the custom CNN model against established pre-trained models (VGG-16, ResNet-50, Inception-v3) using transfer learning.
• To assess the model's effectiveness, accuracy, and computational efficiency on a limited dataset.

Main Methods:

• Implementation of a custom Convolutional Neural Network (CNN) architecture.
• Application of Data Augmentation and Image Processing techniques to enhance the dataset.
• Utilizing transfer learning to compare the custom CNN with pre-trained models: VGG-16, ResNet-50, and Inception-v3.

Main Results:

• The custom CNN model achieved a 100% accuracy rate in classifying brain MRI scans.
• Comparative performance: Custom CNN (100%), VGG-16 (96%), ResNet-50 (89%), and Inception-v3 (75%).
• The developed model demonstrated high effectiveness with very low complexity and reduced computational power requirements.

Conclusions:

• The proposed CNN model offers a highly accurate and computationally efficient solution for brain tumor detection in MRI scans.
• The custom CNN model outperforms popular pre-trained models, suggesting its potential for clinical application in medical diagnostics.
• Further research with larger datasets can validate the robustness and generalizability of this approach for brain tumor classification.