Classifying tumor brain images using parallel deep learning algorithms.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Deep Learning

Background:

• Medical images are crucial for disease diagnosis, particularly for brain conditions.
• Convolution neural networks (CNNs) are powerful deep learning tools for image analysis.
• Accurate and generalizable diagnostic tools are needed to support clinical decision-making.

Purpose of the Study:

• To develop and evaluate a novel deep parallel CNN model for medical image-based disease diagnosis.
• To assess the model's accuracy and generalizability compared to existing methods.
• To determine the potential of the proposed model as a decision support tool for radiologists.

Main Methods:

• A deep parallel CNN model integrating AlexNet and VGGNet architectures was proposed.
• Network layers were structured differently, with features combined and categorized using a softmax function.
• The model was tested on a medical image database (FIGSHARE), ensuring no individual observations from the training set were in the test set for generalizability.

Main Results:

• The proposed model achieved high accuracy: 99.14% for binary classification and 98.78% for multi-class classification on the FIGSHARE database.
• Performance significantly surpassed existing Support Vector Machine (SVM) models.
• The model demonstrated superior results across all evaluation criteria when compared to existing methods.

Conclusions:

• The proposed deep parallel CNN model is highly effective for medical image analysis and disease diagnosis.
• The model exhibits excellent accuracy and generalizability, making it a reliable decision support tool.
• This AI-driven approach shows promise for enhancing diagnostic capabilities in radiology.