摘要
针对光子计数较少导致心肌灌注PET(MP-PET)图像质量严重退化问题,提出基于TGV正则化的MP-PET图像恢复模型。模型结合低秩稀疏分解理论对MP-PET序列图像的目标区域和背景区域进行分离,针对目标区域进行TGV约束从而改善图像质量。分别对仿真数据及临床扫描数据进行实验,采用视觉及不同的量化指标进行恢复效果评价。实验结果表明,提出的TGV模型具有优质的恢复效果。
Spatial resolution of myocardial perfusion （MP） PET images is degenerated due to limited photon counts detected. To achieve resolution - improved images, TGV based dynamic MP - PET images restoration model had been proposed in this work. Proposed model decomposed the dynamic MP - PET images into dynamic and background components incorporating the low rank plus sparse decomposition theory, then applied TGV constraint to improve the image quality. Both simulation study and clinical data experiments demonstrate substantial improvements of our proposed model, both visual and quantitative accuracy.
出处
《核电子学与探测技术》
北大核心
2017年第4期382-387,共6页
Nuclear Electronics & Detection Technology
基金
教育部高等学校博士学科专项科研基金(20134433120017)
广东省科技计划项目(2016A020216016)
广东省医学科研基金(A2016044)
广东食品药品职业学院自然科学研究项目(2016YZ026)资助
关键词
心肌灌注PET显像
广义全变分
图像恢复
myocardial perfusion PET imaging
total generalized variation
image restoration
