|
|
|
|
|
|
Abstract
The quality of MRI images in hospitals often poses a significant challenge due to issues such as noise and blurring during the image acquisition process. Insufficient image resolution can adversely affect the accuracy of image segmentation, necessitating a solution for segmenting low-resolution MRI images effectively. In this study, we propose an ensemble method that combines SR-GAN (Super-Resolution Generative Adversarial Network) and the UBMRI-Seg (Ultra-Blurry MRI Segmentation) method for low-resolution MRI image segmentation. The objective is to achieve comparable segmentation results between low-resolution and high-resolution images. The proposed approach offers several benefits. Firstly, it assists doctors in making accurate diagnoses based on sharper and improved image quality. By enhancing the resolution of MRI images, the method improves the precision of segmentation results. Moreover, it saves time and cost. High-resolution MRI machines typically require longer scan times and incur higher expenses. By utilizing the ensemble method, the need for high-resolution machines can be reduced, resulting in time and cost savings. Experimental evaluations demonstrate the effectiveness of the proposed approach in achieving accurate segmentation results for low-resolution MRI images. The integration of SR-GAN and UBMRI-Seg enhances the image resolution and improves the quality of MRI segmentation. These advancements contribute to more reliable and efficient image-based diagnoses, benefiting both patients and medical practitioners.