Grosser Oliver S, Ruf Juri, Kupitz Dennis, Czuczwara Damian, Loewenthal David, Thormann Markus, Furth Christian, Ricke Jens, Pech Maciej, Kreissl Michael C, Amthauer Holger
Nuklearmedizin. 2018 Aug;57(4):153-159. doi: 10.3413/Nukmed-0953-17-12. Epub 2018 Aug 20.
Low-dose-computed tomography (LD-CT) is used in nuclear medicine hybrid imaging (e.g., SPECT/CT) for attenuation correction of emission data and anatomical correlation of findings. However, there are currently no standards for image quality (e. g., detectability) comparable to those for diagnostic CT. Therefore, the aim of this explorative study was to evaluate retrospective LDCT data in terms of CT image quality and detectability of anatomical structures.
Two readers blindly scored abdominal LD-CT images (n = 40 patients) in terms of detectability (n = 20 structures/patient), image quality, and readers' confidence in scoring the image quality for a clinically hybrid imaging protocol. Results were analysed by ANOVA to identify factors (e. g., anatomical structures) that influenced performance scores. The inter-rater agreement was evaluated by determining the chance-corrected Cohen's Kappa coefficient.
Image noise was acceptable for anatomical correlation in 96.1 % of the readings with an almost perfect inter-rater agreement (K = 0.85). A detectability of at least 80 % was observed in 13/20 (K ≥ 0.7) and 90 % in 9/20 (K ≥ 0.85) of the structures analysed by both readers. The confidence of both readers in scoring image quality was at least sufficient in 98.8 % of the examined patients (K = 0.95).
Although LD-CT protocols commonly used in hybrid imaging have a poor image quality not suitable for primary CT diagnostics, they enable detection of a variety of anatomical structures. LDCT can therefore also be referenced in the associated reports for anatomical correlation of findings from SPECT imaging.
低剂量计算机断层扫描（LD-CT）用于核医学混合成像（如SPECT/CT），以对发射数据进行衰减校正并对检查结果进行解剖学关联。然而，目前尚无与诊断性CT相当的图像质量标准（如可检测性）。因此，本探索性研究的目的是根据CT图像质量和解剖结构的可检测性评估回顾性LDCT数据。
两名阅片者对腹部LD-CT图像（n = 40例患者）进行盲法评分，内容包括可检测性（n = 20个结构/患者）、图像质量以及阅片者对临床混合成像方案图像质量评分的信心。通过方差分析对结果进行分析，以确定影响性能评分的因素（如解剖结构）。通过计算机会校正的Cohen's Kappa系数评估阅片者间的一致性。
在96.1%的读数中，图像噪声对于解剖学关联是可接受的，阅片者间一致性几乎完美（K = 0.85）。两位阅片者分析的结构中，13/20（K≥0.7）的结构可检测性至少为80%，9/20（K≥0.85）的结构可检测性为90%。两位阅片者对图像质量评分的信心在98.8%的受检患者中至少是足够的（K = 0.95）。
尽管混合成像中常用的LD-CT方案图像质量较差，不适用于初级CT诊断，但它们能够检测多种解剖结构。因此，在相关报告中也可引用LDCT对SPECT成像结果进行解剖学关联。
