PET／MR与PET／CT的对比研究

目的通过与PET／CT在病灶检测及定量分析方面的比较，论证PET／MR一体机应用于临床的可行性。方法2012年5月至2013年2月共300例患者同天间隔15—35min行PET／CT和PET／MR检查，其中临床及影像学资料完整的277例男165例，女112例，平均年龄（52．9±12．6）岁]纳入该研究。扫描范围由下颌至大腿根部。PET／CT采集为2或3min／床位，PET／MR采集为5min／床位，MRI图像包括T1、T2及弥散加权成像，图像重建采用Vender软件及MRI散射校正Mu图。PET、CT、MRI中只要有1项为阳性，即认为PET／CT及PET／MR的结果为阳性。勾画ROI并计算SUVmax。不同显像技术间探测效率的比较采用配对X2检验，并对PET／CT与PET／MR的SUVmax进行相关分析。结果277例患者随访证实阳性220例，病灶353个。与PET／CT比，PET／MR少发现了6个病灶，但多发现了30个病灶。两者的探测效率差异有统计学意义（P〈0．05）。以病灶和患者计，两者的一致性分别为89.8％（317／353）和85．9％（189／220）。相同部位病灶PET／CT和PET／MR的SUVmax的相关性良好（rs=0．91，P〈0．01），多数正常组织（除肺和主气管）的SUVmax也具有良好的相关性（rs=0．62-0．76，均P〈0．01）。结论PET／MR一体机在半定量分析方面与PET／CT有良好的相关性，在躯干部尤其是腹、盆腔病变及软组织病变检测方面优于PET／CT，可以为临床提供可靠的影像学诊断。

Comparison between PET/MR and PET/CT in evaluation of oncological patients

Objective To verify the feasibility of the integrated PET/MR for oncological applica- tions by comparing PET/MR with PET/CT in terms of lesion detection and quantitative measurement. Meth- ods A total of 277 patients （ 165 males, 112 females, average age （52.9±12.6） years） voluntarily partici- pated in this same-day PET/CT and PET/MR comparative study. The time interval between the two studies was 15-35 min. PET/CT images were acquired and reconstructed following standard protocols. PET/MR covered the body trunk with a sequence combination of transverse T1 weighted imaging （WI） 3D-volumetric interpolated breath-hold, T2WI turbo spin echo with fat saturation, diffusion-weighted imaging, and simulta- neous PET acquisition. PET images were reconstructed by vender-provided attenuation correction methods. The results of PET/CT and PET/MR were regarded as positive if any modality （ CT, PET or MRI） was pos- itive. SUVm∽ was obtained by the manually drawn ROI. Detection rates were compared with Xz test and SUV∽,∽ from the two modalities was analyzed with Spearman correlation analysis. Results A total of 353 le- sions were detected in 220 patients. Compared to PET/CT, PET/MR revealed 30 additional true-positive le- sions, while missed 6. The detection rates between PET/CT and PET/MR were significantly different （P〈 0. 05）. The lesion-based and patient-based consistency was 89.8% （317/353） and 85.9% （189/220）, re- spectively. There were significant correlations of SUV,,ax between PET/MR and PET/CT for lesions（ r∽ = 0. 91, P〈0.01 ） and for normal tissues（r∽ =0.62-0.76, all P〈0.01）. Conclusions With reference to PET/CT, in- tegrated PET/MR may provide comparable semi-quantitative measurements of pathological lesions as well as normal tissues. Integrated PET/MR may be more effective to detect lesions in abdomen and pelvis.