Usefulness Evaluation of Deep Learning Super Resolution Technique in Clinical Application of Image Guided Diagnosis and Treatment using Magnetic Resonance Imaging of Metastatic Brain Tumor

This study aimed to implement artificial intelligence (AI) techniques using magnetic resonance imaging (MRI) of metastatic brain tumors for potential application in diagnosis and surgical environments. The experimental method involved reconstructing MRI data by applying an AI based super resolution (SR) technique to the ground truth (GT) images obtained using T2 weighted imaging (T2WI), T1 weighted imaging (T1WI), contrast enhanced T1WI (CE-T1WI), and contrast enhanced fluid attenuated inversion recovery (CE-FLAIR). The performance of the reconstructed images was evaluated based on peak signal to noise ratio (PSNR), structural similarity index measure (SSIM), and gamma knife radiosurgery (GKRS) coordinate systems. The results showed that the PSNR and SSIM values of the SR technique were higher than those achieved with conventional image post processing methods, and no differences were observed in the radiosurgery coordinates. Consequently, SR demonstrated superior image quality improvement without altering the coordinate system, confirming its utility as a viable technique for the diagnosis and surgical treatment of metastatic brain tumors.