Jin H, Su Z
University of Oklahoma Health Science Center, Oklahoma City, AA.
University of Florida, Jacksonville, FL.
Med Phys. 2012 Jun;39(6Part13):3755. doi: 10.1118/1.4735288.
Applicability of the AlignRT surface imaging system was extensively investigated for real-time motion tracking in radiation therapy of lung cancer.
A 4D computer-controlled motion phantom was employed to simulate a human breathing motion. An anthropomorphic thoracic phantom (a coronal cross-sectional area: ∼1,200 cm ) was monitored during the 4D motion tracking. A breathing cycle from a healthy subject (breathing frequency: 0.2 Hz; maximum amplitude: vertical 7 mm, longitudinal 6 mm, and lateral 1.2 mm) was monitored and fed into the phantom control system. The real time monitoring was investigated by modifying a region of detection (the whole, a half, and a quarter of the area) and combination of 3 cameras (patient left, middle, and right).
The real-time tracking errors (detected breathing motion - the input of the subject's breathing) of the phantom motion using one camera (lateral, longitudinal, vertical) were (0.02±0.14 mm, -0.23±0.48 mm, -0.10±0.48 mm) with ipsilateral halves of the area (frame rate: 1.52 Hz), (0.06±0.10 mm, -0.30±0.30 mm, -0.03-0.35 mm) with ipsilateral quarters (frame rate: 1.95 Hz), and (0.19±0.96 mm, 0.29±1.50 mm, -0.46±1.58 mm) with contralateral quarters (frame rate: 3.04 Hz). The detection errors using 2-camera combinations (left-middle and right-middle) were (-0.06±0.06 mm, -0.43±0.21 mm, 0.00±0.22 mm) for the ipsilateral quarters and (-0.62±0.18 mm, -1.20±0.37 mm, 0.03±0.30 mm) for the contralateral quarters, respectively. The 3-camera tracking using halves of the area presented an error of (-0.04±0.12 mm, -0.69±0.35 mm, -0.08±0.38 mm) with a frame rate of 0.62 Hz.
To guarantee a sub-millimeter accuracy of realtime motion tracking in the lung treatment, combination of 2 cameras or less, a frame rate of greater than the Nyquist frequency (in this study fN = 2B (breathing frequency) = 0.4 Hz), and a tracking area of less than an ipsilateral half of thoracic area is highly recommended.
广泛研究AlignRT表面成像系统在肺癌放射治疗实时运动跟踪中的适用性。
采用四维计算机控制运动体模模拟人体呼吸运动。在四维运动跟踪过程中监测一个人体胸部体模（冠状截面积：约1200平方厘米）。监测一名健康受试者的呼吸周期（呼吸频率：0.2赫兹；最大幅度：垂直方向7毫米、纵向6毫米、横向1.2毫米）并输入体模控制系统。通过修改检测区域（整个区域、一半区域和四分之一区域）以及三台摄像机（患者左侧、中间和右侧）的组合来研究实时监测情况。
使用一台摄像机（横向、纵向、垂直方向）对体模运动进行实时跟踪误差（检测到的呼吸运动 - 受试者呼吸输入），对于同侧一半区域（帧率：1.52赫兹）分别为(0.02±0.14毫米, -0.23±0.48毫米, -0.10±0.48毫米)，对于同侧四分之一区域（帧率：1.95赫兹）为(0.06±0.10毫米, -0.30±0.30毫米, -0.03 - 0.35毫米)，对于对侧四分之一区域（帧率：3.04赫兹）为(0.19±0.96毫米, 0.29±1.50毫米, -0.46±1.58毫米)。使用两台摄像机组合（左 - 中、右 - 中）时，对于同侧四分之一区域的检测误差分别为(-0.06±0.06毫米, -0.43±0.21毫米, 0.00±0.22毫米)，对于对侧四分之一区域为(-0.62±0.18毫米, -1.20±0.37毫米, 0.03±0.30毫米)。使用区域一半的三台摄像机跟踪时，帧率为0.62赫兹，误差为(-0.04±0.12毫米, -0.69±0.35毫米, -0.08±0.38毫米)。
为确保肺部治疗中实时运动跟踪的亚毫米精度，强烈建议使用两台或更少摄像机组合、帧率大于奈奎斯特频率（本研究中fN = 2B（呼吸频率） = 0.4赫兹）以及跟踪区域小于胸部同侧一半区域。
