Enhancing Image Quality of Knee MRI in Low-Field MRI Using Deep Learning [Presidential Award Proceedings]

Abstract

　This study employed deep learning techniques to enhance the image quality of low-field knee MRI, with the goal of aligning it with high-field standards and visualizing transformation effects via texture analysis. Through training on 3T data (51 individuals) and by validating on 0.4T data (25 individuals) with two distinct models (super-resolution and contrast generation), the results might reveal an enhancement in the image quality of low-field knee MRI.