不同滤波方法对脑部PET图像质量和SUV值的影响

Impact of Different Filter Method on Brain PET Image Quality and Standard Uptake Value

目的探讨不同滤波方法对^(68)Ga-1,4,7,10-四氮杂环十二烷-1,4,7,10-四乙酸-1-柰丙氨酸-奥曲肽(DOTA-NOC)脑部PET图像质量和标准摄取值(Standard Uptake Value,SUV)的影响。方法选用美国国家电气制造者协会(NEMA)标准模体和行^(68)Ga-DOTA-NOC PET检查的39例患者作为研究对象,模体和脑部PET图像重建均采用有序子集最大期望法(OSEM)+飞行时间技术(TOF)+点扩散技术(PSF),滤波方法采用None、Smooth1、Smooth2、Smooth3、Enhance1和Enhance2,图像质量评估采用模体变异系数(CV)值、对比度、信噪比、垂体SUVpeak、SUVmax、SUVmean和SUVsd。结果基于Smooth3的模体和脑部PET图像质量均最优。Smooth3对应的模体CV值(16.57%)和对比度(41.16)均最小,对应的信噪比(242.32)最大。基于Smooth3的SUV值小于其他滤波方法,且不同滤波方法所得垂体SUVpeak、SUVmax、SUVmean和SUVsd值得分差异之间均无统计学意义(F=0.7354,P=0.5976;F=1.166,P=0.3270;F=1.725,P=0.1299;F=1.160,P=0.3301)。结论不同滤波函数获得的^(68)Ga-DOTA-NOC脑部PET图像质量和垂体SUV值差异较大,平滑函数Smooth3能有效抑制PET图像噪声,持续获得高质量脑部PET图像,为PET重建算法提供参考依据。

Objective To investigate the effect of different filter method on ^68Ga-DOTA-1-Nal3-octreotide (^68Ga-DOTA-NOC) brain PET image quality and standard uptake value (SUV). Methods The body phantom of National Electrical Manufacturers Association and International Electro technical Commission and 39 consecutive patients who underwent clinical ^68Ga-DOTA-NOC were selected as research object. The phantom and brain PET data were reconstructed with the baseline ordered subsets expectation maximization algorithm (OSEM)+point spread function (PSF)+time of flight (TOF) model. PET image quality was evaluated by the coefficient of variance, contrast and signal-to-noise ratio in phantom study and pituitary SUVpeak, SUVmax, SUVmean and SUVsd in clinical study. Results The bets quality of phantom and clinical brain PET image was achieved by Smooth3. The CV and contrast reaches the minimum value at 16.57 % and 41.16 respectively, and the maximum of SNR was 242.32 based on Smooth3. In clinical study, the SUVs of Smooth3 were lower than other filter methods, and results of one-way ANOVA shows pituitary SUVpeak, SUVmax, SUVmean and SUVsd for 6 all filter method groups had no significant statistical difference (F=0.7354, P=0.5976;F=1.166, P=0.3270;F=1.725, P=0.1299;F=1.160, P=0.3301). Conclusion ^68Ga-DOTA-NOC Brain PET image and pituitary SUV from different filter method has differences, smooth3 can suppress the noise effectively and consistently obtain good quality PET image, which can offer reference for brain PET image reconstruction algorithm.