TumorDetNet: A unified deep learning model for brain tumor detection and classification.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Medicine
• Neuro-oncology

Background:

• Early and accurate brain tumor diagnosis is critical for effective treatment and patient survival.
• Magnetic resonance imaging (MRI) is a non-invasive tool for tumor identification, but manual analysis is time-consuming and challenging.
• Computer-aided diagnostics (CAD) systems offer a solution to automate and improve the accuracy of brain tumor detection and classification.

Purpose of the Study:

• To propose a unified, end-to-end deep learning model named TumorDetNet for brain tumor detection and classification.
• To evaluate the performance of TumorDetNet on standard Kaggle brain tumor MRI datasets.
• To demonstrate the model's capability in classifying tumor types and malignancy.

Main Methods:

• Developed TumorDetNet, a deep learning framework utilizing 48 convolution layers with Leaky ReLU (LReLU) and ReLU activation functions.
• Incorporated average pooling and a dropout layer to learn distinctive patterns and prevent overfitting.
• Employed fully connected and softmax layers for final tumor detection and multi-class classification.

Main Results:

• TumorDetNet achieved 99.83% accuracy in brain tumor detection across six Kaggle datasets.
• The model demonstrated 100% accuracy in classifying tumors as benign or malignant.
• Classification accuracy for specific tumor types (glioma, pituitary, meningioma) reached 99.27%.

Conclusions:

• The proposed TumorDetNet model shows significant potential for reliable and accurate brain tumor identification and classification.
• This deep learning approach can enhance computer-aided diagnostics systems, improving diagnostic efficiency and patient care.
• The high accuracy achieved highlights the effectiveness of the developed methodology in addressing the challenges of brain tumor analysis.