From the *Institute of Clinical Radiology and Nuclear Medicine, University Medical Center Mannheim, Heidelberg University, Heidelberg, Germany; †Department of Radiology, Peking University First Hospital, Beijing, China; ‡Siemens Healthcare, Erlangen, Germany; §Department of Radiology, New York University, New York City, NY; and ∥Institut für Radiologie, Kantonsspital Baden, Baden, Switzerland.
Invest Radiol. 2017 Dec;52(12):715-724. doi: 10.1097/RLI.0000000000000397.
Purpose of this study was to compare the quality of perfusion maps obtained from prototypical free-breathing magnetic resonance imaging (MRI) with continuous golden-angle radial sampling and iterative reconstruction (GRASP) to conventional acquisition using time-resolved angiography with interleaved stochastic trajectories (TWIST) in patients with rectal cancer.
Forty cases were included for retrospective analysis. Twenty of the patients received routine multiparametric MRI at 3 T for rectal cancer staging, including perfusion measurement with GRASP or TWIST (10 patients for each technique, including 5 prechemoradiation and 5 postchemoradiation). Twenty patients without history of rectal disease served as control group (10 GRASP, 10 TWIST). GRASP images were reconstructed at temporal resolution of 3.45 seconds (21 spokes/frame). A voxel-by-voxel deconvolution approach was used to determine rectal plasma flow (mL/100 mL per minute). Regions of interest were placed at 3 levels within the tumor and normal rectum (lower, middle, and upper part). The quality of morphologic images, perfusion maps, and arterial input function were scored by 2 blinded radiologists. Independent t tests were applied.
Three patients of the TWIST control group had to be excluded due to technical failure of the sequence. Significantly higher scores for the perfusion maps and arterial input functions (total cohort) were obtained using GRASP (P < 0.05). Artifacts in the perfusion maps were rated significantly lower than for TWIST (P < 0.05). In the healthy rectum cohort, the average plasma flow of normal rectal wall was 31.78 ± 7.39 mL/100 mL per minute with GRASP, compared with 77.62 ± 34.08 mL/100 mL per minute with TWIST, indicating much lower variance for GRASP. Plasma flow values obtained with both methods enabled distinguishing between normal rectal wall and rectal cancer, both before and after chemoradiation. Morphologic image quality was generally higher with GRASP (P < 0.01).
GRASP perfusion imaging can distinguish between normal rectum and rectal cancers with higher image quality and less variance than TWIST. Additional morphologic assessment with high spatial resolution from the GRASP acquisition may increase the accuracy and diagnostic confidence of the examination.
本研究的目的是比较源自原型自由呼吸磁共振成像（MRI）的灌注图的质量，这种成像采用连续黄金角度径向采样和迭代重建（GRASP），与直肠癌患者使用时间分辨血管造影与交错随机轨迹（TWIST）的常规采集进行比较。
对 40 例患者进行回顾性分析。其中 20 例患者接受 3T 多参数 MRI 常规直肠癌症分期检查，包括使用 GRASP 或 TWIST 进行灌注测量（每种技术各 10 例，包括 5 例放化疗前和 5 例放化疗后）。20 例无直肠疾病史的患者作为对照组（10 例采用 GRASP，10 例采用 TWIST）。GRASP 图像在 3.45 秒的时间分辨率下重建（21 个 spokes/帧）。采用体素分解方法确定直肠血浆流量（每分钟每 100 毫升毫升）。在肿瘤和正常直肠的 3 个水平放置感兴趣区域（下部、中部和上部）。2 名盲法放射科医生对形态图像、灌注图和动脉输入功能的质量进行评分。应用独立 t 检验。
由于序列技术故障，TWIST 对照组的 3 例患者不得不被排除在外。GRASP 获得的灌注图和动脉输入功能评分显著更高（总队列，P < 0.05）。灌注图中的伪影评分显著低于 TWIST（P < 0.05）。在健康直肠队列中，GRASP 下正常直肠壁的平均血浆流量为 31.78 ± 7.39 mL/100 mL/分钟，而 TWIST 下为 77.62 ± 34.08 mL/100 mL/分钟，表明 GRASP 的方差明显较低。放化疗前后，两种方法均能区分正常直肠壁和直肠癌，且均能获得血浆流量值。GRASP 的形态图像质量通常更高（P < 0.01）。
GRASP 灌注成像可以比 TWIST 更清晰地区分正常直肠和直肠癌，并且具有更高的图像质量和更低的方差。GRASP 采集的高空间分辨率可提供额外的形态学评估，从而提高检查的准确性和诊断信心。
