PET/CT医学图像去噪方法的研究  

Medical image denoising method based on PET/CT

作　　者：强彦[1] 卢军佐[1] 赵涓涓[1] 

机构地区：[1]太原理工大学计算机科学与技术学院,太原030024

出　　处：《清华大学学报（自然科学版）》2012年第8期1056-1060,共5页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金资助项目(60970059);山西省自然科学基金资助项目(2012011015-1);山西省科技攻关项目(20120313032-3)

摘　　要：噪声元素在光子量有限的最大似然期望最大化(MLEM)方法重建PET/CT(positron emission tomography/computed tomography)图像研究的过程中影响甚至加重图像特征明暗度变化。为减小这类影响,该文提出了一种包含解剖端信息的中值非局部平均算法(AMNLM)。该算法是对适用低信噪比图像去噪的非局部平均算法(NLM)的改进。将该算法应用于临床肺部病变患者PET/CT图像的分析研究。参数反演结果表明,在视觉评估、病变对比度和噪声的权衡方面与中值滤波方法、小波滤波方法和传统NLM方法相比较,该方法具有更好的去噪效果,从而也验证了该方法的有效性。The intensity gradient due to noise elements may compete with or even exceed that in the images in the research of photon limited maximum likelihood expectation maximization （MLEM） based positron emission tomography/eomputed tomography （PET/CT） image reconstruction. An average median non local means （AMNLM）method was developed to decrease the interference of the noise elements during image processing. The AMNLM method is based on the traditional image denoising method of the median non local means filter, incorporating with anatomical side information. The method was used to analyze the clinical pulmonary disease PET/CT images. The parameters obtained from experiments show that the AMNLM method outperforms the anisotropic-median diffusion filtering （AMDF） and non local means（NLM）in terms of visual assessment and trade off between lesion contrast and noise, which shows the validity of this method.

关 键 词：医学图像去噪 PET/CT 非局部平均算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]