文章摘要

文章信息/Info

Title:: Prior constrained regularized level set for automatic ?kidney region of interest extraction

作者:: 魏文涛1; 郑秀娟1△; 宋少莉2; 苏敏1; 1.四川大学电气信息学院自动化系，四川成都 610065；2.上海交通大学医学院附属仁济医院核医学科，上海 200127

Author(s):: WEI Wentao1; ZHENG Xiujuan1; SONG Shaoli2; SU Min1; 1.The College of Electrical Engineering and Information Technology, Sichuan University, Chengdu 610065, China；2.Department of Nuclear Medicine, Renji Hospital, School of Medicine, Shanghai Jiaotong University,Shanghai 200127，China

Keywords:: Regularized level set; Priori information; Kidney region of interest; Gates method; Glomerular filtration rate

摘要:: 临床上常采用肾动态图联合Gates分析法来获得肾小球滤过率（GFR）并用于肾功能评价，但该方法需要手动勾画肾脏及背景感兴趣区（ROI）进行定量计算，其结果易受到图像质量、医生经验等因素的影响，且耗时耗力。为了快速且客观准确地获取GFR值，本研究提出了先验信息约束的正则化水平集肾脏感兴趣区自动提取方法。该方法以正则化水平集算法为基础，融入肾动态图先验信息进行水平集演化约束，实现肾脏ROI自动提取。由临床数据验证，该自动方法能够准确且稳定地提取肾脏及背景ROI，实现准确的GFR计算，尤其有利于肾功能严重受损时的GFR值估计。

Abstract:: As the crucial indicator of renal function, glomerular filtration rate (GFR) can be calculated by Gates method with dynamic kidney single photon emission computed tomography (SPCET) imaging. In this method, it is necessary to delineate kidney region of interest (ROI) and corresponding background manually for quantitative calculation. This method is subjective and labor-intensive. Moreover, the estimates are easily affected by the image qualities and clinicians′ experience. In this paper, a priori information constrained regularized level set was proposed to automatically extract precious kidney ROIs to achieve the accurate and objective GFR estimations. Inherited from the regularized level set method, the proposed method used priori information to constrain the level set curve evolution, then achieved automatic extraction of kidney ROI.The results of clinical applications shows that this method can obtain accurate and stable kidney ROIs and GFR estimates especially for the severely low functioningkidneys.