Xu Jennifer, Sisniega Alejandro, Zbijewski Wojciech, Dang Hao, Stayman J Webster, Wang Xiaohui, Foos David H, Aygun Nafi, Koliatsos Vassillis E, Siewerdsen Jeffrey H
Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD 21205, USA.
Phys Med Biol. 2016 Aug 21;61(16):5973-92. doi: 10.1088/0031-9155/61/16/5973. Epub 2016 Jul 20.
The effects of detector readout gain mode and bowtie filters on cone-beam CT (CBCT) image quality and dose were characterized for a new CBCT system developed for point-of-care imaging of the head, with potential application to diagnosis of traumatic brain injury, intracranial hemorrhage (ICH), and stroke. A detector performance model was extended to include the effects of detector readout gain on electronic digitization noise. The noise performance for high-gain (HG), low-gain (LG), and dual-gain (DG) detector readout was evaluated, and the benefit associated with HG mode in regions free from detector saturation was quantified. Such benefit could be realized (without detector saturation) either via DG mode or by incorporation of a bowtie filter. Therefore, three bowtie filters were investigated that varied in thickness and curvature. A polyenergetic gain correction method was developed to equalize the detector response between the flood-field and projection data in the presence of a bowtie. The effect of bowtie filters on dose, scatter-to-primary ratio, contrast, and noise was quantified in phantom studies, and results were compared to a high-speed Monte Carlo (MC) simulation to characterize x-ray scatter and dose distributions in the head. Imaging in DG mode improved the contrast-to-noise ratio (CNR) by ~15% compared to LG mode at a dose (D 0, measured at the center of a 16 cm CTDI phantom) of 19 mGy. MC dose calculations agreed with CTDI measurements and showed that bowtie filters reduce peripheral dose by as much as 50% at the same central dose. Bowtie filters were found to increase the CNR per unit square-root dose near the center of the image by ~5-20% depending on bowtie thickness, but reduced CNR in the periphery by ~10-40%. Images acquired at equal CTDIw with and without a bowtie demonstrated a 24% increase in CNR at the center of an anthropomorphic head phantom. Combining a thick bowtie filter with a short arc (180° + fan angle) scan centered on the posterior of the head reduced dose to the eye lens by up to 90%. Acquisition in DG mode (without a bowtie filter) was beneficial to the detection of small, low contrast lesions (e.g. subtle ICH) in CBCT. While bowtie filters were found to reduce dose, mitigate sensor saturation at the periphery in HG mode, and improve CNR at the center of the image, the image quality at the periphery was slightly reduced compared to DG mode, and the use of a bowtie required careful implementation of the polyenergetic flood-field correction to avoid artifacts.
针对一种为头部即时成像开发的新型锥束CT（CBCT）系统，研究了探测器读出增益模式和蝴蝶结滤波器对CBCT图像质量及剂量的影响，该系统在创伤性脑损伤、颅内出血（ICH）和中风的诊断方面具有潜在应用价值。扩展了探测器性能模型，以纳入探测器读出增益对电子数字化噪声的影响。评估了高增益（HG）、低增益（LG）和双增益（DG）探测器读出的噪声性能，并对在无探测器饱和区域中HG模式的益处进行了量化。这种益处（在无探测器饱和情况下）可通过DG模式或采用蝴蝶结滤波器来实现。因此，研究了三种厚度和曲率不同的蝴蝶结滤波器。开发了一种多能增益校正方法，以在存在蝴蝶结的情况下使探测器在泛射场和投影数据之间的响应均衡。在体模研究中对蝴蝶结滤波器对剂量、散射与原发射线比率、对比度和噪声的影响进行了量化，并将结果与高速蒙特卡罗（MC）模拟进行比较，以表征头部的X射线散射和剂量分布。在剂量（在16 cm CTDI体模中心测量的D0）为19 mGy时，与LG模式相比，DG模式成像将对比度噪声比（CNR）提高了约15%。MC剂量计算与CTDI测量结果一致，表明在相同中心剂量下，蝴蝶结滤波器可将周边剂量降低多达50%。发现蝴蝶结滤波器根据其厚度可使图像中心每单位平方根剂量的CNR提高约5 - 20%，但使周边的CNR降低约10 - 40%。在有和无蝴蝶结的情况下以相等的CTDIw采集的图像显示，在仿人头体模中心CNR提高了24%。将厚蝴蝶结滤波器与以头部后部为中心的短弧（180° + 扇形角）扫描相结合，可使晶状体剂量降低多达90%。在DG模式下采集（无蝴蝶结滤波器）有利于在CBCT中检测小的、低对比度病变（如细微ICH）。虽然发现蝴蝶结滤波器可降低剂量、减轻HG模式下周边的传感器饱和并提高图像中心的CNR，但与DG模式相比，周边的图像质量略有降低，并且使用蝴蝶结需要谨慎实施多能泛射场校正以避免伪影。
