ACRF Image X Institute, Faculty of Medicine and Health, The University of Sydney, Australia.
Phys Med Biol. 2020 Apr 8;65(7):075012. doi: 10.1088/1361-6560/ab7abe.
Breathing variations during 4D CT imaging often manifest as geometric irregularities known as respiratory-induced image artifacts and ultimately effect radiotherapy treatment efficacy. To reduce such image artifacts we developed Respiratory Adaptive Computed Tomography (REACT) to trigger CT acquisition during periods of regular breathing. For the first time, we integrate REACT with clinical hardware and hypothesize that REACT will reduce respiratory-induced image artifacts ≥ 4 mm compared to conventional 4D CT. 4D image sets were acquired using REACT and conventional 4D CT on a Siemens Somatom scanner. Scans were taken for 13 respiratory traces (12 patients) that were reproduced on a lung-motion phantom. Motion was observed by the Varian RPM system and sent to the REACT software where breathing irregularity was evaluated in real-time and used to trigger the imaging beam. REACT and conventional 4D CT images were compared to a ground truth static-phantom image and compared for absolute geometric differences within the region-of-interest. Breathing irregularity during imaging was retrospectively assessed using the root-mean-square error of the RPM measured respiratory signal during beam on (RMSE_Beam_on) for each phase of the respiratory cycle. REACT significantly reduced the average frequency of respiratory-induced image artifacts ≥ 4 mm by 70% for the tumor (p = 0.003) and 76% for the lung (p = 0.0002) compared to conventional 4D CT. Volume reductions of 10% to 6% of the tumor and 2% to 1% of the lung compared to conventional 4D CT were seen. Breathing irregularity during imaging (RMSE_Beam_on) was significantly reduced by 27% (p = 0.013) using the REACT method. For the first time, REACT was successfully integrated with clinical hardware. Our findings support the hypothesis that REACT significantly reduced respiratory-induced image artifacts compared to conventional 4D CT. These experimental results provide compelling evidence for further REACT investigation, potentially providing clearer images for clinical use.
4D CT 成像期间的呼吸变化通常表现为几何不规则性，称为呼吸引起的图像伪影，最终影响放射治疗效果。为了减少这种图像伪影，我们开发了呼吸自适应计算机断层扫描（REACT），以便在有规律的呼吸期间触发 CT 采集。这是首次将 REACT 与临床硬件相结合，并假设与常规 4D CT 相比，REACT 将减少≥4mm 的呼吸引起的图像伪影。使用 Siemens Somatom 扫描仪上的 REACT 和常规 4D CT 采集 4D 图像集。对 12 名患者的 13 个呼吸轨迹进行了扫描，并在肺运动体模上进行了复制。通过 Varian RPM 系统观察运动，并将其发送到 REACT 软件，该软件实时评估呼吸不规则性，并用于触发成像束。将 REACT 和常规 4D CT 图像与地面真实静态体模图像进行比较，并比较感兴趣区域内的绝对几何差异。使用 RPM 测量的呼吸信号在束上的均方根误差（RMSE_Beam_on），对每个呼吸周期阶段的成像期间的呼吸不规则性进行回顾性评估。与常规 4D CT 相比，REACT 显著降低了肿瘤（p=0.003）和肺部（p=0.0002）≥4mm 的呼吸引起的图像伪影的平均频率，分别降低了 70%和 76%。与常规 4D CT 相比，肿瘤体积减少了 10%至 6%，肺部体积减少了 2%至 1%。使用 REACT 方法，成像期间的呼吸不规则性（RMSE_Beam_on）显著降低了 27%（p=0.013）。这是首次成功地将 REACT 与临床硬件相结合。我们的研究结果支持这样的假设，即与常规 4D CT 相比，REACT 显著降低了呼吸引起的图像伪影。这些实验结果为进一步的 REACT 研究提供了有力的证据，可能为临床应用提供更清晰的图像。
