Cone beam CT-based set-up strategies with and without rotational correction for stereotactic body radiation therapy in the liver

<b>Background:</b> Accurate patient positioning is crucial in stereotactic body radiation therapy (SBRT) due to a high dose regimen. Cone-beam computed tomography (CBCT) is often used for patient positioning based on radio-opaque markers. We compared six CBCT-based set-up strategies with or without rotational correction. <b>Material and methods:</b> Twenty-nine patients with three implanted markers received 3-6 fraction liver SBRT. The markers were delineated on the mid-ventilation phase of a 4D-planning-CT. One pretreatment CBCT was acquired per fraction. Set-up strategy 1 used only translational correction based on manual marker match between the CBCT and planning CT. Set-up strategy 2 used automatic 6 degrees-of-freedom registration of the vertebrae closest to the target. The 3D marker trajectories were also extracted from the projections and the mean position of each marker was calculated and used for set-up strategies 3–6. Translational correction only was used for strategy 3. Translational and rotational corrections were used for strategies 4–6 with the rotation being either vertebrae based (strategy 4), or marker based and constrained to ±3° (strategy 5) or unconstrained (strategy 6). The resulting set-up error was calculated as the 3D root-mean-square set-up error of the three markers. The set-up error of the spinal cord was calculated for all strategies. <b>Results:</b> The bony anatomy set-up (2) had the largest set-up error (5.8 mm). The marker-based set-up with unconstrained rotations (6) had the smallest set-up error (0.8 mm) but the largest spinal cord set-up error (12.1 mm). The marker-based set-up with translational correction only (3) or with bony anatomy rotational correction (4) had equivalent set-up error (1.3 mm) but rotational correction reduced the spinal cord set-up error from 4.1 mm to 3.5 mm. <b>Conclusions:</b> Marker-based set-up was substantially better than bony-anatomy set-up. Rotational correction may improve the set-up, but further investigations are required to determine the optimal correction strategy.

<b>背景：</b> 立体定向体部放射治疗（stereotactic body radiation therapy, SBRT）采用高剂量给药方案，因此患者体位的精准摆位至关重要。锥形束计算机断层扫描（cone-beam computed tomography, CBCT）常借助不透射线标记物完成患者体位摆位。本研究对比了六种基于CBCT的摆位策略，分别包含或不包含旋转校正。
<b>材料与方法：</b> 本研究纳入29例植入3枚标记物的患者，均接受3~6分割疗程的肝脏SBRT治疗。标记物在4D计划CT的中期通气相上完成勾画。每个分割疗程均采集1次治疗前CBCT图像。摆位策略1仅通过CBCT与计划CT的手动标记物匹配实施平移校正。摆位策略2采用与靶区毗邻椎体的自动6自由度配准。研究同时从投影数据中提取3枚标记物的三维运动轨迹，计算每枚标记物的平均位置，并将其应用于摆位策略3~6。策略3仅采用平移校正；策略4至6则同时采用平移与旋转校正，其中旋转校正分别基于椎体（策略4）、基于标记物且约束于±3°（策略5），或无约束（策略6）。最终摆位误差以3枚标记物的三维均方根摆位误差进行计算，并针对所有策略计算脊髓的摆位误差。
<b>结果：</b> 基于骨解剖结构的摆位（策略2）的摆位误差最大，达5.8 mm。仅采用无约束旋转校正的标记物基摆位（策略6）的摆位误差最小，为0.8 mm，但脊髓摆位误差最大，达12.1 mm。仅采用平移校正的标记物基摆位（策略3）与基于椎体旋转校正的摆位（策略4）的摆位误差相当，均为1.3 mm；而旋转校正可将脊髓摆位误差从4.1 mm降至3.5 mm。
<b>结论：</b> 基于标记物的摆位效果显著优于基于骨解剖结构的摆位。旋转校正可改善摆位质量，但仍需开展进一步研究以确定最优的校正策略。