摘要
为了提出一种单光子发射断层扫描成像(SPECT)影像呼吸运动参数估计和伪影校正方案,该文通过对伪影图像的方向微分和自相关分析分别辨识出体模的运动方向和幅度,根据匀速运动和正弦振动拟合的系统点扩散函数,运用Wiener滤波和Richardson-Lucy盲反卷积(RL)算法校正伪影。结果表明:5个典型运动方向和幅度辨识的结果均与实际值接近,RL算法校正伪影经过5次迭代后图像质量总体优于维纳滤波。通过图像后处理估计运动参数,运用迭代反卷积可以有效的校正图像伪影,可能成为一种更方便实用SPECT/PET(正电子发射断层扫描成像)图像呼吸伪影解决方案。
Respiratory motion identification and corrections for the motion artifacts in SPECT imaging were analyzed in a phantom study by identifying,the blur direction in the SPECT image through a directional derivative analysis of the motion blurred image.The blur extent was estimated by an auto-correlation analysis of the pseudo-whitened blurred image.Then,a Wiener filter and the Richardson-Lucy blind deconvolution algorithm（RL） were employed to restore the motion blurred image according to the estimated point spread function.The results show that the identification results for 5 typical simulated motions are very close to the true value and that the RL algorithm enables better motion artifact correction within 5 iterations than the Wiener filter.This method can be used to identify respiratory motion with a post-reconstruction based deconvolution algorithm as a promising method in clinical practice to correct motion artifacts in SPECT/PET images.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第11期1896-1900,共5页
Journal of Tsinghua University(Science and Technology)
基金
广东省自然科学基金资助项目(No.2008B030303055)
