Hussain Fahad Ahmed, Mail Noor, Shamy Abdulrahman M, Suliman Alghamdi, Saoudi Abdelhamid
King Abdulaziz Medical City.
J Appl Clin Med Phys. 2016 May 8;17(3):419-432. doi: 10.1120/jacmp.v17i3.5903.
Image quality is a key issue in radiology, particularly in a clinical setting where it is important to achieve accurate diagnoses while minimizing radiation dose. Some computed tomography (CT) manufacturers have introduced algorithms that claim significant dose reduction. In this study, we assessed CT image quality produced by two reconstruction algorithms provided with GE Healthcare's Discovery 690 Elite positron emission tomography (PET) CT scanner. Image quality was measured for images obtained at various doses with both conventional filtered back-projection (FBP) and adaptive statistical iterative reconstruction (ASIR) algorithms. A stan-dard CT dose index (CTDI) phantom and a pencil ionization chamber were used to measure the CT dose at 120 kVp and an exposure of 260 mAs. Image quality was assessed using two phantoms. CT images of both phantoms were acquired at tube voltage (kV) of 120 with exposures ranging from 25 mAs to 400 mAs. Images were reconstructed using FBP and ASIR ranging from 10% to 100%, then analyzed for noise, low-contrast detectability, contrast-to-noise ratio (CNR), and modulation transfer function (MTF). Noise was 4.6 HU in water phantom images acquired at 260 mAs/FBP 120 kV and 130 mAs/50% ASIR 120 kV. The large objects (fre-quency < 7 lp/cm) retained fairly acceptable image quality at 130 mAs/50% ASIR, compared to 260 mAs/FBP. The application of ASIR for small objects (frequency >7 lp/cm) showed poor visibility compared to FBP at 260 mAs and even worse for images acquired at less than 130 mAs. ASIR blending more than 50% at low dose tends to reduce contrast of small objects (frequency >7 lp/cm). We concluded that dose reduction and ASIR should be applied with close attention if the objects to be detected or diagnosed are small (frequency > 7 lp/cm). Further investigations are required to correlate the small objects (frequency > 7 lp/cm) to patient anatomy and clinical diagnosis.
图像质量是放射学中的一个关键问题，尤其是在临床环境中，在将辐射剂量降至最低的同时实现准确诊断非常重要。一些计算机断层扫描（CT）制造商推出了声称能显著降低剂量的算法。在本研究中，我们评估了通用电气医疗集团的Discovery 690 Elite正电子发射断层扫描（PET）CT扫描仪所配备的两种重建算法产生的CT图像质量。使用传统的滤波反投影（FBP）算法和自适应统计迭代重建（ASIR）算法，对在不同剂量下获得的图像的图像质量进行了测量。使用标准CT剂量指数（CTDI）模体和铅笔电离室在120 kVp和260 mAs的曝光条件下测量CT剂量。使用两种模体评估图像质量。两种模体的CT图像均在120 kV的管电压下采集，曝光范围为25 mAs至400 mAs。使用10%至100%的FBP和ASIR对图像进行重建，然后分析图像的噪声、低对比度可探测性、对比度噪声比（CNR）和调制传递函数（MTF）。在260 mAs/FBP 120 kV和130 mAs/50% ASIR 120 kV采集的水模体图像中，噪声为4.6 HU。与260 mAs/FBP相比，大物体（频率<7 lp/cm）在130 mAs/50% ASIR时保持了相当可接受的图像质量。与260 mAs时的FBP相比，ASIR在小物体（频率>7 lp/cm）上的应用显示出较差的可视性，对于在低于130 mAs采集的图像甚至更差。在低剂量下，ASIR混合超过50%往往会降低小物体（频率>7 lp/cm）的对比度。我们得出结论，如果要检测或诊断的物体较小（频率>7 lp/cm），则应密切关注剂量降低和ASIR的应用。需要进一步研究将小物体（频率>7 lp/cm）与患者解剖结构和临床诊断相关联。
