A deep learning model to classify and detect brain abnormalities in portable microwave based imaging system.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Biomedical Engineering

Background:

• Automated detection of brain abnormalities is crucial for medical diagnosis and monitoring.
• Reconstructed microwave (RMW) imaging offers a portable solution for brain imaging.

Purpose of the Study:

• To develop and evaluate a deep learning model for automatic classification and detection of brain tumors in RMW images.
• To assess the performance of the YOLOv5 object detection model in a portable microwave head imaging system (MWHI).

Main Methods:

• Collected 400 RMW images of non-tumor and tumor samples from an MWHI system.
• Applied image pre-processing and augmentation to create a dataset of 4400 images.
• Trained and tested YOLOv5s, YOLOv5m, and YOLOv5l models, with 80% for training and 20% for testing.

Main Results:

• The YOLOv5l model demonstrated superior performance over YOLOv5s, YOLOv5m, and other state-of-the-art models.
• Achieved 96.32% accuracy, 95.17% precision, 94.98% sensitivity, 95.28% specificity, 95.53% F1-score, and 96.12% mAP.
• The YOLOv5l model accurately detected tumors with bounding boxes and classified them as benign or malignant.

Conclusions:

• The YOLOv5l object detection model is reliable for automatic tumor detection and classification in RMW images.
• This approach shows potential for real-time applications in portable microwave brain imaging.
• The study highlights the effectiveness of deep learning in analyzing medical imaging data for neurological abnormalities.