Li Xiang, Zeng Chen, Li Yige, Liu Huan, Liu Ling, Zeng Wenbing, Yang Ran
From the Department of Radiology, Chongqing University Three Gorges Hospital, Chongqing.
Department of Endocrinology, Harbin Medical University, Heilongjiang.
J Comput Assist Tomogr. 2023;47(3):500-506. doi: 10.1097/RCT.0000000000001421. Epub 2023 Mar 3.
The aim of this study was to compare 3 computed tomography perfusion (CTP) software packages in the estimation of infarct core volumes, hypoperfusion volumes, and mismatch volumes.
Forty-three patients with large vessel occlusion in the anterior circulation who underwent CTP imaging were postprocessed by 3 software packages: RAPID, advantage workstation (AW), and NovoStroke Kit (NSK). Infarct core volumes and hypoperfusion volumes were generated by RAPID with default settings. The AW and NSK threshold settings were the following: infarct core (cerebral blood flow [CBF] <8 mL/min/100 g, CBF <10 mL/min/100 g, CBF <12 mL/min/100 g, and cerebral blood volume [CBV] <1 mL/100 g) and hypoperfusion (T max >6 seconds). Mismatch volumes were then obtained for all the combinations of the settings. Bland-Altman, intraclass correlation coefficient (ICC), and Spearman ρ or Pearson correlation coefficient were applied for statistical analysis.
In the estimation of infarct core volumes, good agreement was observed between AW and RAPID when CBV <1 mL/100 g (ICC, 0.767; P < 0.001). For hypoperfusion volumes, good agreement (ICC, 0.811; P < 0.001) and strong correlation ( r = 0.856; P < 0.001) were observed between NSK and RAPID. For mismatch volumes, the setting of CBF <10 mL/min/100 g combined with hypoperfusion with NSK resulted in moderate agreement (ICC, 0.699; P < 0.001) with RAPID, which was the best among all other settings.
The estimation results varied among different software packages. Advantage workstation had the best agreement with RAPID in the estimation of infarct core volumes when CBV <1 mL/100 g. NovoStroke Kit had better agreement and correlation with RAPID in the estimation of hypoperfusion volumes. NovoStroke Kit also had moderate agreement with RAPID in estimating mismatch volumes.
本研究旨在比较3种计算机断层扫描灌注（CTP）软件包在梗死核心体积、低灌注体积和不匹配体积估计方面的表现。
对43例接受CTP成像的前循环大血管闭塞患者，使用3种软件包进行后处理：RAPID、优势工作站（AW）和NovoStroke套件（NSK）。梗死核心体积和低灌注体积由RAPID采用默认设置生成。AW和NSK的阈值设置如下：梗死核心（脑血流量[CBF]<8 mL/min/100 g、CBF<10 mL/min/100 g、CBF<12 mL/min/100 g，以及脑血容量[CBV]<1 mL/100 g）和低灌注（Tmax>6秒）。然后针对所有设置组合获取不匹配体积。采用Bland-Altman分析、组内相关系数（ICC）以及Spearman ρ或Pearson相关系数进行统计分析。
在梗死核心体积估计方面，当CBV<1 mL/100 g时，AW与RAPID之间观察到良好的一致性（ICC，0.767；P<0.001）。对于低灌注体积，NSK与RAPID之间观察到良好的一致性（ICC，0.811；P<0.001）和强相关性（r = 0.856；P<0.001）。对于不匹配体积，NSK的CBF<10 mL/min/100 g与低灌注设置相结合，与RAPID产生了中度一致性（ICC，0.699；P<0.001），这在所有其他设置中是最好的。
不同软件包的估计结果存在差异。当CBV<1 mL/100 g时，优势工作站在梗死核心体积估计方面与RAPID的一致性最佳。NovoStroke套件在低灌注体积估计方面与RAPID具有更好的一致性和相关性。NovoStroke套件在估计不匹配体积方面与RAPID也具有中度一致性。
