Brain Tumor Segmentation and Classification from Sensor-Based Portable Microwave Brain Imaging System Using...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Biomedical Engineering

Background:

• Manual brain tumor segmentation and classification are time-consuming and critical for disease monitoring.
• Reconstructed microwave (RMW) imaging offers a potential modality for brain imaging.
• Automated analysis of RMW images is needed to improve diagnostic efficiency.

Purpose of the Study:

• To develop and evaluate novel deep learning models for automated brain tumor segmentation and RMW image classification.
• To introduce MicrowaveSegNet (MSegNet) for tumor segmentation and BrainImageNet (BINet) for image classification.
• To assess the performance of these models against state-of-the-art methods.

Main Methods:

• A dataset of 300 RMW brain images was collected and augmented to 6000 images per fold for 5-fold cross-validation.
• Developed a lightweight segmentation model, MicrowaveSegNet (MSegNet).
• Developed a classification model, BrainImageNet (BINet), for RMW images.

Main Results:

• MSegNet achieved an Intersection-over-Union (IoU) of 86.92% and a Dice score of 93.10% for tumor segmentation.
• BINet achieved high accuracy (89.33% raw, 98.33% segmented) for three-class classification of RMW images.
• The proposed cascaded models demonstrated strong performance in segmenting tumors and classifying RMW images.

Conclusions:

• The developed MSegNet and BINet models provide an effective automated solution for brain tumor segmentation and RMW image classification.
• The proposed models show significant potential for integration into sensors-based microwave brain imaging (SMBI) systems.
• Automated analysis using MSegNet and BINet can enhance the investigation and monitoring of brain diseases.