深度学习在磁共振图像超分辨率重建中的应用  

Application of deep learning in super-resolution reconstruction of magnetic resonance images

作　　者：于会昌 刘士远 YU Huichang;LIU Shiyuan(School of Health Science and Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China;Department of Diagnostic Radiology,Shanghai Changzheng Hospital,Naval Medical University,Shanghai 200003,China)

机构地区：[1]上海理工大学健康科学与工程学院,上海200093 [2]海军军医大学附属长征医院放射诊断科,上海200003

出　　处：《中国医学物理学杂志》2024年第10期1243-1248,共6页Chinese Journal of Medical Physics

基　　金：国家自然科学基金(81930049)。

摘　　要：磁共振成像(MRI)是医学影像学中一项重要的非侵入性检查技术,受限于磁共振硬件设备和扫描时间,有些磁共振图像具有较低的空间分辨率;深度学习技术的兴起为解决MRI图像分辨率问题提供了新的途径。本研究首先概述了MRI图像超分辨率重建的背景;其次,深入探讨了在MRI图像超分辨率重建任务中,各种深度学习方法的应用,并对这些方法进行详细的分析,对每种算法的工作原理、优势及其在图像重建过程中的效能表现进行评估;最后,讨论了深度学习技术在MRI图像超分辨率重建中的关键挑战,并对未来研究趋势进行展望。Magnetic resonance imaging(MRI)is a significant non-invasive diagnostic technique in medical imaging.Due to limitations in MRI hardware and scanning time,some MRI images have relatively low spatial resolution.The rise of deep learning technology offers a new approach to improve the resolution of MRI images.The study outlines the background of MRI super-resolution reconstruction,delves into the applications of various deep learning methods in MRI super-resolution reconstruction and offers a detailed analysis of these methods,evaluating their working principles,advantages,and performance efficiency in image reconstruction.Additionally,it also discusses the key challenges of deep learning technology in MRI super-resolution reconstruction,and provides prospects for future research trends.

关 键 词：磁共振成像 超分辨率重建 深度学习 神经网络 综述 

分 类 号：R318[医药卫生—生物医学工程]