Nosrati Reyhaneh, Song William Y, Wronski Matthew, Pejović-Milić Ana, Morton Gerard, Stanisz Greg J
Department of Physics, Ryerson University, Toronto, Canada; Physical Sciences Platform, Sunnybrook Research Institute, Toronto, Canada.
Department of Radiation Oncology, Virginia Commonwealth University, Richmond, VA.
Brachytherapy. 2019 Nov-Dec;18(6):863-874. doi: 10.1016/j.brachy.2019.06.004. Epub 2019 Jul 20.
The lack of positive contrast from brachytherapy seeds in conventional MR images remains a major challenge toward an MRI-only workflow for postimplant dosimetry of low-dose-rate brachytherapy. In this work, the feasibility of our recently proposed MRI-only workflow in clinically relevant scenarios is investigated and the necessary modifications in image acquisition and processing pipeline are proposed for transition to the clinic.
Four prostate phantoms with a total of 321 I-125 implanted dummy seeds and three patients with a total of 168 implanted seeds were scanned using a gradient echo sequence on 1.5 T and 3T MR scanners. Quantitative susceptibility mapping (QSM) was performed for seed visualization. Before QSM, the seed-induced distortion correction was performed followed by edge enhancement. Seed localization was performed using spatial clustering algorithms and was compared with CT. In addition, feasibility of the proposed method on detection of prostatic calcifications was studied.
The proposed susceptibility-based algorithm generated consistent positive contrast for the seeds in phantoms and patients. All the 321 seeds in the four phantoms were correctly identified; the MR-derived seeds centroids agreed well with CT-derived positions (average error = 0.5 ± 0.3 mm). The proposed algorithm for seed visualization was found to be orientation invariant. In patient cases, all seeds were visualized and correctly localized (average error = 1.2 ± 0.9 mm); no significant differences between dose volume histogram parameters were found. Prostatic calcifications were depicted with negative contrast on QSM and spatially agreed with CT.
The proposed MRI-based approach has great potential to replace the current CT-based practices. Additional patient studies are necessary to further optimize and validate the workflow.
在传统磁共振图像中，近距离放射治疗种子缺乏正性对比，这仍然是低剂量率近距离放射治疗植入后剂量测定仅使用磁共振成像工作流程面临的主要挑战。在这项工作中，研究了我们最近提出的仅使用磁共振成像工作流程在临床相关场景中的可行性，并提出了图像采集和处理流程中向临床过渡所需的修改。
使用1.5T和3T磁共振扫描仪上的梯度回波序列对四个含有总共321个I-125植入模拟种子的前列腺模型以及三名总共植入168颗种子的患者进行扫描。进行定量磁化率映射（QSM）以实现种子可视化。在进行QSM之前，先进行种子诱导畸变校正，然后进行边缘增强。使用空间聚类算法进行种子定位，并与CT进行比较。此外，还研究了所提出方法在检测前列腺钙化方面的可行性。
所提出的基于磁化率的算法在模型和患者中为种子生成了一致的正性对比。四个模型中的所有321颗种子均被正确识别；磁共振衍生的种子质心与CT衍生的位置吻合良好（平均误差 = 0.5 ± 0.3毫米）。发现所提出的种子可视化算法具有方向不变性。在患者病例中，所有种子均被可视化并正确定位（平均误差 = 1.2 ± 0.9毫米）；剂量体积直方图参数之间未发现显著差异。前列腺钙化在QSM上以负性对比显示，并且在空间上与CT一致。
所提出的基于磁共振成像的方法有很大潜力取代当前基于CT的做法。需要进一步开展患者研究以进一步优化和验证该工作流程。
