Brain tumor detection and segmentation: Interactive framework with a visual interface and feedback facility for...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Oncology
• Neurosurgery

Background:

• Malignant brain tumors are highly fatal due to challenges in early detection.
• Current diagnostic methods include invasive and non-invasive techniques, with medical imaging being key.
• Magnetic Resonance Imaging (MRI) is a primary non-invasive tool for brain tumor assessment.

Purpose of the Study:

• To develop an automated system for detecting and segmenting brain tumors from 2D and 3D brain MRIs.
• To create a user-friendly web application for accessing and analyzing brain MRI data.
• To improve the accuracy and efficiency of brain tumor diagnosis through advanced AI models.

Main Methods:

• Utilized deep neural networks, including Convolutional Neural Networks (CNN), U-Net, and U-Net++.
• Developed a web application interface for user interaction and data input.
• Implemented automated detection and segmentation algorithms for brain tumors in MRI scans.

Main Results:

• Achieved over 90% accuracy in brain tumor detection and segmentation.
• Demonstrated high performance with Dice scores exceeding 90%.
• The system allows users to upload MRIs or access hospital databases for analysis.

Conclusions:

• The automated web application system effectively detects and segments brain tumors from MRIs.
• The system's high accuracy and user-friendly interface support early diagnosis and treatment planning.
• Incorporating healthcare professional feedback enhances model training for future improvements.