Poulsen Per Rugaard, Jonassen Johnny, Schmidt Mai Lykkegaard, Jensen Carsten
Department of Oncology, Aarhus University Hospital, 8000 Aarhus C, Denmark and Department of Clinical Medicine, Aarhus University, 8000 Aarhus C, Denmark.
Department of Medical Physics, Aarhus University Hospital, 8000 Aarhus C, Denmark.
Med Phys. 2015 Nov;42(11):6549-57. doi: 10.1118/1.4933248.
The gantry-mounted kilovoltage (kV) imager of modern linear accelerators can be used for real-time tumor localization during radiation treatment delivery. However, the kV image quality often suffers from cross-scatter from the megavoltage (MV) treatment beam. This study investigates readout of unexposed kV frames as a means to improve the kV image quality in a series of experiments and a theoretical model of the observed image quality improvements.
A series of fluoroscopic images were acquired of a solid water phantom with an embedded gold marker and an air cavity with and without simultaneous radiation of the phantom with a 6 MV beam delivered perpendicular to the kV beam with 300 and 600 monitor units per minute (MU/min). An in-house built device triggered readout of zero, one, or multiple unexposed frames between the kV exposures. The unexposed frames contained part of the MV scatter, consequently reducing the amount of MV scatter accumulated in the exposed frames. The image quality with and without unexposed frame readout was quantified as the contrast-to-noise ratio (CNR) of the gold marker and air cavity for a range of imaging frequencies from 1 to 15 Hz. To gain more insight into the observed CNR changes, the image lag of the kV imager was measured and used as input in a simple model that describes the CNR with unexposed frame readout in terms of the contrast, kV noise, and MV noise measured without readout of unexposed frames.
Without readout of unexposed kV frames, the quality of intratreatment kV images decreased dramatically with reduced kV frequencies due to MV scatter. The gold marker was only visible for imaging frequencies≥3 Hz at 300 MU/min and ≥5 Hz for 600 MU/min. Visibility of the air cavity required even higher imaging frequencies. Readout of multiple unexposed frames ensured visibility of both structures at all imaging frequencies and a CNR that was independent of the kV frame rate. The image lag was 12.2%, 2.2%, and 0.9% in the first, second, and third frame after an exposure. The CNR model predicted the CNR with triggered image readout with a mean absolute error of 2.0% for the gold marker.
A device that triggers readout of unexposed frames during kV fluoroscopy was built and shown to greatly improve the quality of intratreatment kV images. A simple theoretical model successfully described the CNR improvements with the device.
现代直线加速器的机架式千伏（kV）成像仪可用于放射治疗过程中的实时肿瘤定位。然而，kV图像质量常常受到兆伏（MV）治疗束交叉散射的影响。本研究通过一系列实验和对观察到的图像质量改善的理论模型，研究未曝光kV帧的读出作为提高kV图像质量的一种方法。
对一个带有嵌入式金标记物的固体水模体和一个气腔进行了一系列透视图像采集，分别在有和没有用6MV束垂直于kV束以每分钟300和600监测单位（MU/min）同时对模体进行照射的情况下进行。一个内部构建的装置触发在kV曝光之间读出零帧、一帧或多帧未曝光帧。未曝光帧包含部分MV散射，从而减少了累积在曝光帧中的MV散射量。对于1至15Hz范围内的一系列成像频率，将有无未曝光帧读出时的图像质量量化为金标记物和气腔的对比度噪声比（CNR）。为了更深入了解观察到的CNR变化，测量了kV成像仪的图像滞后，并将其用作一个简单模型的输入，该模型根据在未读出未曝光帧时测量的对比度、kV噪声和MV噪声来描述有未曝光帧读出时的CNR。
在未读出未曝光kV帧的情况下，由于MV散射，治疗过程中kV图像的质量随着kV频率降低而显著下降。在300MU/min时，金标记物仅在成像频率≥3Hz时可见，在600MU/min时≥5Hz时可见。气腔的可见性需要更高的成像频率。读出多帧未曝光帧可确保在所有成像频率下两种结构均可见，且CNR与kV帧率无关。曝光后第一帧、第二帧和第三帧的图像滞后分别为12.2%、2.2%和0.9%。CNR模型预测触发图像读出时金标记物的CNR，平均绝对误差为2.0%。
构建了一种在kV透视过程中触发未曝光帧读出的装置，结果表明该装置可显著提高治疗过程中kV图像的质量。一个简单的理论模型成功地描述了该装置带来的CNR改善情况。
