AbstractBackground: Approximately more than 25% of MRI examinations are Brain MRI. One of the important Brain MRI sequences is FLAIR Fatsat, but the image results of FLAIR Fatsat Brain MRI sequences have noise and long scanning time. To overcome this, the parallel imaging technique GRAPPA and denoising convolutional neural network (CNN) post processing can be used so as to reduce scanning time and reduce noise in the image results of the FLAIR Fatsat Brain MRI sequence.
Objective: Knowing the difference in image quality between before and after the application of CNN denoising technique on MRI examination using parallel imaging GRAPPA on axial T2 FLAIR Fatsat Brain MRI.
Methods: This study uses retrospective data by collecting 3362 axial T2 Flair Fatsat GRAPPA Brain MRI images, with a sample size of 92 images, comparing the original image with the denoising image using CNN and assessing the Peak Signal Noise Ratio (PSNR), Structural Similarity Index (SSIM), Signal to Noise Ratio (SNR) and Contrast to Noise Ratio (CNR).
Results: The research proves that of the 92 samples obtained, the performance of the CNN denoising technique has a difference in image quality between before and after the application of the CNN denoising technique, with an SNR value of 13.40, CNR value of 24.95, PSNR value of 32.62, SSIM value of 0.82.
Conclusion: CNN denoising technique can be considered as post processing of MRI Brain image quality improvement, there is a difference in the value of SNR, CNR, and obtained good average PSNR and SSIM values indicating the quality of the resulting image after the application of CNN denoising technique is getting better.