Classification of Brain Tumor Images Using CNN.

Area of Science:

• Medical Imaging and Diagnostics
• Artificial Intelligence in Healthcare
• Machine Learning for Oncology

Background:

• Brain tumors are malignant conditions characterized by uncontrolled cell growth.
• Deep learning, particularly Convolutional Neural Networks (CNNs), shows promise in medical image analysis.
• Accurate and efficient diagnostic tools are crucial for timely brain tumor detection.

Purpose of the Study:

• To develop and evaluate a CNN model for classifying brain MRI scans.
• To compare the performance of a custom-trained CNN against a pretrained VGG-16 model using transfer learning.
• To assess the accuracy and computational efficiency of the proposed CNN model.

Main Methods:

• Brain MRI images were categorized using a custom Convolutional Neural Network (CNN).
• Data augmentation and image processing techniques were employed to enhance the dataset.
• The performance of the scratch CNN model was compared with a pretrained VGG-16 model via transfer learning.

Main Results:

• The developed CNN model achieved 100% accuracy in classifying brain MRI scans.
• The custom CNN model demonstrated significantly lower complexity rates compared to VGG-16.
• The scratch CNN model outperformed the pretrained VGG-16 model (100% vs. 96% accuracy).

Conclusions:

• The proposed CNN model offers a highly accurate and computationally efficient solution for brain tumor classification from MRI data.
• This approach surpasses existing pretrained methods in terms of accuracy and resource utilization.
• The findings suggest the potential of custom CNNs for improving diagnostic capabilities in neuro-oncology.