Multiple Brain Tumor Classification with Dense CNN Architecture Using Brain MRI Images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurology

Background:

• Brain Magnetic Resonance (MR) imaging is crucial for diagnosing neurological conditions like tumors, strokes, and dementia.
• Deep learning, particularly Convolutional Neural Networks (CNNs), offers advanced capabilities for medical image analysis.
• Existing transfer learning methods have limitations in brain MRI classification.

Purpose of the Study:

• To develop a novel modular deep learning model for brain MRI classification.
• To leverage and improve upon established transfer learning techniques (DenseNet, VGG16, basic CNNs).
• To enhance the accuracy of detecting abnormalities in brain MR images.

Main Methods:

• A new modular deep learning architecture was designed.
• The model was trained and tested using open-source brain tumor MRI images from Kaggle.
• Training utilized both an 80/20 data split and 10-fold cross-validation.

Main Results:

• The proposed deep learning model demonstrated improved classification performance compared to existing transfer learning methods.
• The model effectively classified brain MR images, showing enhanced diagnostic potential.
• An increase in processing time was noted alongside performance gains.

Conclusions:

• The novel modular deep learning model offers a promising advancement in brain MRI analysis for disease detection.
• This approach effectively balances improved classification accuracy with computational demands.
• Further research can optimize processing time for clinical application.