Automated Brain Tumor Detection Using Convolutional Neural Network

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuro-oncology

Background:

• Accurate brain tumor detection is critical for timely medical intervention.
• Conventional methods rely on manual radiological analysis, prone to errors and variability.
• Deep learning offers potential to improve diagnostic accuracy and efficiency.

Purpose of the Study:

• To evaluate the efficacy of U-Net and Single-Shot Multibox Detector (SSD) deep learning models for early brain tumor detection.
• To compare the performance of U-Net for segmentation and SSD for object detection in brain tumor identification.
• To assess the potential of these AI techniques in enhancing neuro-oncology diagnostics.

Main Methods:

• Utilized U-Net, a convolutional neural network (CNN) renowned for medical image segmentation.
• Employed Single-Shot Multibox Detector (SSD), an established object detection algorithm.
• Applied these models to medical scans for brain tumor identification and localization.

Main Results:

• The U-Net model demonstrated high performance, achieving 97.73% accuracy in brain tumor segmentation.
• The SSD model achieved 58% accuracy, indicating potential as a supplementary tool.
• U-Net showed exceptional precision in identifying and localizing brain tumors.

Conclusions:

• U-Net is a highly effective method for precise brain tumor detection in medical imaging.
• Deep learning, particularly U-Net, significantly improves early detection outcomes in neuro-oncology.
• Further research can explore enhancing diagnostic accuracy with these advanced AI techniques.