深度学习在核素心肌灌注显像中的研究进展  

Research progress of deep learning in nuclear myocardial perfusion imaging

作　　者：宋昊 武志芳[1] 柴象飞 郗锐 葛浩 李思进[1] Song Hao;Wu Zhifang;Chai Xiangfei;Xi Rui;Ge Hao;Li Sijin(Department of Nuclear Medicine,First Hospital of Shanxi Medical University/Collaborative Innovation Center of Molecular Imaging Precision Diagnosis and Treatment,Taiyuan 030001,China)

机构地区：[1]山西医科大学第一医院核医学科、分子影像精准诊疗省部共建协同创新中心,太原030001

出　　处：《中华核医学与分子影像杂志》2024年第2期116-119,共4页Chinese Journal of Nuclear Medicine and Molecular Imaging

基　　金：国家自然科学基金(82027804,81971655)。

摘　　要：近年来,以深度学习(DL)为代表的人工智能(AI)技术快速发展,智慧医疗已成为AI最重要的应用领域之一。核素心肌灌注显像(MPI)作为评估心肌血流最准确的无创检查,具有重要的临床价值。目前,利用DL算法基于MPI图像建立学习模型仍处于研究阶段,实现广泛推广还需开展更多的外部验证及迭代更新。该文拟对DL算法在MPI中的应用进行综述,以期为进一步的研究提供思路和方法学参照。In recent years,artificial intelligence(AI)technology represented by deep learning(DL)has developed rapidly,and smart medical care has become one of the most important application areas of AI.As the most accurate noninvasive test to assess myocardial blood flow,myocardial perfusion imaging(MPI)has important clinical values.At present,the use of DL algorithms to establish learning models for MPI images is still in the research stage,and more external verification and iterative updates are needed before it can be widely used in real time clinical practice.In this article,the application of DL algorithms in MPI is comprehensively elaborated to provide a basis and direction for further research.

关 键 词：心肌灌注显像 深度学习 发展趋势 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] R817.4[自动化与计算机技术—控制科学与工程]