Enhancing brain tumor detection in MRI images using YOLO-NeuroBoost model

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Brain tumors require early detection and precise localization in MRI for effective diagnosis and treatment.
• Current methods face challenges in accurately identifying and locating brain tumors within MRI scans.

Purpose of the Study:

• To develop an advanced model for accurate brain tumor detection and localization in MRI images.
• To improve the timeliness and effectiveness of brain tumor diagnosis and treatment through enhanced image analysis.

Main Methods:

• Proposed the YOLO-NeuroBoost model, integrating an improved YOLOv8 algorithm.
• Incorporated dynamic convolution KernelWarehouse, Convolutional Block Attention Module (CBAM), and Inner-GIoU loss function.
• Evaluated the model on the Br35H and Roboflow datasets.

Main Results:

• Achieved a mean Average Precision (mAP) of 99.48% on the Br35H dataset.
• Achieved a mAP of 97.71% on the Roboflow dataset.
• Demonstrated high accuracy and efficiency in detecting brain tumors in MRI images.

Conclusions:

• The YOLO-NeuroBoost model significantly improves brain tumor detection accuracy in MRI.
• This advancement offers potential for earlier diagnosis and more effective treatment strategies.
• The research contributes to the progress of medical image analysis in neuro-oncology.