A Novel Generative Adversarial Network-Based Approach for Automated Brain Tumour Segmentation.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Medical image segmentation is complex due to image density.
• Brain tumors are a leading cause of mortality.
• Distinguishing tumorous from non-tumorous cells is challenging.

Purpose of the Study:

• To propose a deep convolutional generative adversarial network (DCGAN) for brain tumor segmentation.
• To address limitations of traditional convolutional neural networks in medical imaging.
• To improve the detection of anomalies in diverse cell types within medical images.

Main Methods:

• A DCGAN model comprising a generator and a discriminator was developed.
• The network focuses on tumor localization, noise reduction, and handling data disparities.
• Brain Magnetic Resonance Imaging (MRI) images were used for segmentation.

Main Results:

• The model achieved a Dice Score Coefficient (DSC) of 0.894, PSNR of 62.084 dB, and SSIM of 0.88912.
• Achieved 97% accuracy with a reduced loss of 0.012.
• Demonstrated successful segmentation of tumorous and benign tissues.

Conclusions:

• The proposed DCGAN approach effectively segments brain tumors in MRI scans.
• This novel method enhances the accuracy of medical image analysis for brain tumor detection.
• The findings contribute to developing advanced tools for cancer diagnosis and treatment planning.