-
选取磁共振直线加速器 Unity治疗的结直肠癌肝转移的患者,并排除存在放疗绝对禁忌症的患者。患者定位采用西门子CT模拟定位机进行四维CT(4D-CT)图像扫描。采取仰卧位、手臂上举放于头部两侧支撑架上等体位完成CT和MR定位阶段扫描。而后基于扫描图像完成靶区和周围危及器官勾勒,其中GTV(Gross Tumor Volume)是影像学所见的原发肿瘤区,内靶区ITV(Internal Target Volume)是参考影像判断由于呼吸运动、膀胱充盈度、胃肠道蠕动等生理活动引起肿瘤形状、位置大小发生改变的靶区范围,ITV外放5~10 mm得到的区域为计划靶区PTV(Planning Target Volume)。目前常规GTV到PTV的外放边界取决于病变的放射学特征和临床判断,并通过4D-CT考虑分次内的不确定性以及摆位误差。放疗计划采用Monaco V5.4系统制定调强放射治疗(IMRT),射野个数为8~9个。采用磁共振加速器治疗过程中,患者按照常规方案即采用腹部加压带限制腹部运动,并用现有2D cine MR成像持续扫描监督肿瘤位置,以保证剂量输送的准确性。患者在每次疗后,首先使用腹部加压带限制腹部运动(RAM组)扫描3.5D MRI图像,随后让患者处于自由呼吸状态(FB组),扫描3.5D MRI图像。
该成像是在正交核磁共振成像实时采集模式的基础上发展起来的。不同于现有2D cine MRI,该方法所呈现图像的中心完全与射野中心重合,依据肿瘤质心和放疗射野特异性方向生成的三个正交成像平面:一是沿着射线方向的两个射束路径观(Beam Path View, BPV)平面;另一个是垂直于射线方向的一个射束方向观(Beam Eye View, BEV)平面。结合所有射野角度的三个正交层面的磁共振实时成像,该方法可提供涉及射野穿过的靶区和周围危及器官位置变化的准确信息,完成后无需后期处理,可直接提供放疗中最关注层面的图像和重点放疗信息的方法。表1列出了采集3.5D MRI序列的关键参数。
参数名称 参数值 参数名称 参数值 Sequence technique 2D Gradient Echo Sequence Number of dynamics in each gantry 150 Contrast Balanced oversampling 50 mm on both sides The scan time per frame/s 0.23 Gradient mode Default Time per gantry (per dynamic)/s 34.7 Fat suppression No Field strength B0/T 1.5 Halfscan No Relaxation time TR/ms 3.6 SENSE 3 Echo time TE/ms 1.82 Uniformity Clear Bandwidth/ Hz 868.7 Reconstruction matrix 320 Flip angle 60 Reconstruction mode Real-time Field of view 300 mm×300 mm Stacks 3 Fold-over suppression Oversampling Slices in each stack 1 Acquired voxel size 2.3 mm×2.26 mm×6 mm Slice orientations Orthogonal to each other Peripheral nerve stimulation(PNS) mode Moderate Coils used MRL_Posterior and MRL_Anterior 由于一个呼吸周期的长度平均为(3.6±0.8) s,因此在一个特定的射野方向上,每个二维动态的采集时间为 34.7 s,可完全覆盖约10个呼吸周期。研究中的肿瘤运动幅度的变化采用肿瘤附近可替代的解剖结构,如:门静脉、肝叶等。由于3.5D MRI图像中肿瘤边界不明显, 而肿瘤附近的血管显像突出并有明确边界。Jupitz 等[15]的研究证明,肝脏肿瘤的运动与肝脏血管之间存在正相关,血管是用于研究肝脏内肿瘤运动的合适替代物。Rosenbery等[16]已经在一项多机构磁共振引导大分割放疗研究中证实采用肿瘤附近的血管等来追踪原发性肝脏或肝转移病灶该方法的有效性。肿瘤附近可替代的血管的勾画步骤如下:1) 通过同一个角度方向下的同一3.5D MRI成像平面按帧与帧采集时间顺序进行排序;2) 由主管医生在同一个角度方向下的同一成像平面 3.5D MRI 成像集上对第一帧上肿瘤附近的血管进行勾画;3) 找到逐帧之间的变形矢量场(DVF),由采用局部自适应正则化的Demons图像配准算法生成DVF [17],血管轮廓由第一帧通过DVF传递到随后的帧;4) 由主管医生对勾画血管轮廓逐帧验证和修正。每个患者的肿瘤替代物的位置在第i帧中采用二维径向矢量Ri表示,每一帧中肿瘤替代物的振幅是△Ri = Ri−Rmean,其中Rmean是同一个角度方向下的同一3.5D MRI成像集上肿瘤替代物振幅的平均值,用95%置信度下的量化分析。对两组之间的差异进行统计分析,首先用Shapiro-Wilk检验判断差异是否符合正态分布,P>0.05则认为差异的分布符合正态分布。当差异符合正态分布时,可使用配对样本t检验;当差异不符合正态分布,则选择Wilcoxon 检验。
A Novel High-resolution Fast Magnetic Resonance Imaging Method Incorporating BEV/BPV Fusion Information
doi: 10.11804/NuclPhysRev.41.2023CNPC72
- Received Date: 2023-08-07
- Rev Recd Date: 2023-10-14
-
Key words:
- High-resolution fast magnetic resonance imaging /
- BEV/BPV information fusion /
- amplitude of motion comparison /
- patients with liver metastases /
- magnetic resonance-guided radiotherapy
Abstract: This study establishes a novel high-resolution fast magnetic resonance imaging(MRI) method that incorporates Beam Eye View(BEV) and Beam Path View(BPV) fusion information. Three liver metastasis patients undergoing MRI guided radiotherapy(MRgRT) were selected. A total of 31 200 frames of MRI images were acquired from each patient using two motion patterns: restricted abdominal motion using an abdominal compression belt(RAM group) and free breathing(FB group). Tumor tracking was performed using nearby vessels with clear boundaries, and the radial vector motion amplitude difference(∆R95) within the 95% confidence interval was calculated. The differences in ΔR95 between the RAM and FB groups in all fractions on the BEV/BPV plane were as follows: for Patient 1, they were all less than 0.58 mm; for Patient 2, they were greater than 2.57 mm; for Patient 3, they were 0.71 mm and 1.05 mm, respectively. The results indicate that the abdominal compression technique can effectively reduce tumor motion magnitude, and the tumor motion magnitude ΔR95 variation is highly individual-specific. This method can serve as an imaging basis for the tumor margin reduction in MRgRT.
Citation: | Jiayun CHEN, Xiaoqi WANG, Shirui QIN, Ran WEI, Deqi CHEN, Ke ZHANG, Ying CAO, Jianrong DAI, yuan TANG. A Novel High-resolution Fast Magnetic Resonance Imaging Method Incorporating BEV/BPV Fusion Information[J]. Nuclear Physics Review. doi: 10.11804/NuclPhysRev.41.2023CNPC72 |