Peteani Giulia, Paganelli Chiara, Giovannelli Anna Chiara, Bachtiary Barbara, Safai Sairos, Rogers Susanne, Pusterla Orso, Riesterer Oliver, Weber Damien Charles, Lomax Antony John, Baroni Guido, Fattori Giovanni
Department of Electronics, Information and Bioengineering, Politecnico di Milano, Milan, Italy.
Center for Proton Therapy, Paul Scherrer Institut, Villigen, Switzerland.
Phys Imaging Radiat Oncol. 2023 Dec 27;29:100529. doi: 10.1016/j.phro.2023.100529. eCollection 2024 Jan.
Imaging of respiration-induced anatomical changes is essential to ensure high accuracy in radiotherapy of lung cancer. We expanded here on methods for retrospective reconstruction of time-resolved volumetric magnetic resonance (4DMR) of the thoracic region and benchmarked the results against 4D computed tomography (4DCT).
MR data of six lung cancer patients were collected by interleaving cine-navigator images with 2D data frame images, acquired across the thorax. The data frame images have been stacked in volumes based on a similarity metric that considers the anatomical deformation of lungs, while addressing ambiguities in respiratory phase detection and interpolation of missing data. The resulting images were validated against cine-navigator images and compared to paired 4DCTs in terms of amplitude and period of motion, assessing differences in internal target volume (ITV) margin definition.
4DMR-based motion amplitude was on average within 1.8 mm of that measured in the corresponding 2D cine-navigator images. In our dataset, the 4DCT motion and the 4DMR median amplitude were always within 3.8 mm. The median period was generally close to CT references, although deviations up to 24 % have been observed. These changes were reflected in the ITV, which was generally larger for MRI than for 4DCT (up to 39.7 %).
The proposed algorithm for retrospective reconstruction of time-resolved volumetric MR provided quality anatomical images with high temporal resolution for motion modelling and treatment planning. The potential for imaging organ motion variability makes 4DMR a valuable complement to standard 4DCT imaging.
呼吸诱导的解剖结构变化成像对于确保肺癌放疗的高精度至关重要。我们在此扩展了胸部区域时间分辨容积磁共振（4DMR）的回顾性重建方法，并将结果与4D计算机断层扫描（4DCT）进行了对比。
通过将电影导航图像与二维数据帧图像交错采集，获取了6例肺癌患者的胸部MR数据。基于考虑肺部解剖变形的相似性度量，将数据帧图像堆叠成容积，同时解决呼吸相位检测和缺失数据插值中的模糊性问题。所得图像与电影导航图像进行了验证，并在运动幅度和周期方面与配对的4DCT进行了比较，评估内部靶区（ITV）边界定义的差异。
基于4DMR的运动幅度平均比相应二维电影导航图像中测得的幅度小1.8毫米。在我们的数据集中，4DCT运动和4DMR中位数幅度始终在3.8毫米以内。中位数周期通常接近CT参考值，尽管观察到高达24%的偏差。这些变化反映在ITV中，MRI的ITV通常比4DCT大（高达39.7%）。
所提出的时间分辨容积MR回顾性重建算法为运动建模和治疗计划提供了具有高时间分辨率的高质量解剖图像。成像器官运动变异性的潜力使4DMR成为标准4DCT成像的有价值补充。
