Reference-Free PRFS MR-Thermometry Using Near-Harmonic 2-D Reconstruction of the Background Phase
Reference-Free PRFS MR-Thermometry Using Near-Harmonic 2-D Reconstruction of the Background Phase
复制标题
DOI：
10.1109/tmi.2011.2168421
复制
发表时间：
2012-02-01
影响因子：
10.6
通讯作者：
Gross, Patrick
中科院分区：
文献类型：
--
作者：
Salomir, Rares;Viallon, Magalie;Gross, Patrick
质子共振频移（PRFS）磁共振测温（MRT）通常是监测热烧蚀的首选方法，通常采用梯度回波（GRE）序列。标准PRFS MRT基于时间参考相位图的相减，因此对组织运动（包括变形）和磁场的外部扰动本质上是敏感的。Rieke等人先前已经描述了无参考（或无参考）PRFS MRT，该方法基于对加热区域外的相位数据进行的二维多项式拟合，以估计感兴趣区域内的背景相位。虽然他们的方法在抗组织运动和磁扰动方面无疑是一个基本的进步，但潜在的数学形式需要一个厚的非加热边界，并且可能受到高阶多项式的数值不稳定性的影响。本文描述了一种新的无参考PRFS MRT方法，该方法使用物理一致的形式，利用均匀或近均匀介质中磁场的数学特性。目前的实现要求MR GRE相位值沿着一个薄的、几乎封闭的和不加热的边界作为输入。这是一个经典狄利克雷问题的二维限制，在每片的基础上工作。通过与“ground truth”数据的比较，该方法已被实验验证，被认为是静态离体组织的标准PRFS方法。在健康志愿者的肝脏中以快速采集（300 ms/图像）对梯度回声相位基线进行“零测量”。在磁共振引导的高强度聚焦超声（MRgHIFU）超声过程中获得的羊肝脏体内数据进行了后处理，以证明其在治疗场景中的适用性。Bland和Altman的平均绝对差值在0.069和0.968之间，而固有的“白”噪声SD为0.23。该方法在志愿者肝脏中的准确度和精密度平均分别为0.13和0.65，固有“白”噪声SD平均为0.51。该方法成功地应用于内径达6.2 cm的大型roi，在c++环境下，每片的计算时间系统地小于100 ms。目前无参考PRFS测温的局限性主要来自于需要提供一个接近封闭的边界，其中MR相位是无伪影的，组织是未加热的，再加上在呼吸过程中可能需要重新定位该边界，以跟踪被监测的解剖区域的运动。本文描述并评价了一种基于谐波函数理论框架的无参考PRFS测温方法。计算时间与局部热疗时的在线监测兼容。目前的无参考MRT方法扩展了工作流程的灵活性，消除了对呼吸触发的需要，实现了更高的时间分辨率，并且对组织的独特事件运动不敏感。
Proton resonance frequency shift (PRFS) MR thermometry (MRT) is the generally preferred method for monitoring thermal ablation, typically implemented with gradient-echo (GRE) sequences. Standard PRFS MRT is based on the subtraction of a temporal reference phase map and is, therefore, intrinsically sensitive to tissue motion (including deformation) and to external perturbation of the magnetic field. Reference-free (or reference-less) PRFS MRT has been previously described by Rieke et al. and was based on a 2-D polynomial fit performed on phase data from outside the heated region, to estimate the background phase inside the region of interest. While their approach was undeniably a fundamental progress in terms of robustness against tissue motion and magnetic perturbations, the underlying mathematical formalism requires a thick unheated border and may be subject to numerical instabilities with high order polynomials. A novel method of reference-free PRFS MRT is described here, using a physically consistent formalism, which exploits mathematical properties of the magnetic field in a homogeneous or near-homogeneous medium.The present implementation requires as input the MR GRE phase values along a thin, nearly-closed and unheated border. This is a 2-D restriction of a classic Dirichlet problem, working on a slice per slice basis. The method has been validated experimentally by comparison with the "ground truth" data, considered to be the standard PRFS method for static ex vivo tissue. "Zero measurement" of the gradient-echo phase baseline was performed in healthy volunteer liver with rapid acquisition (300 ms/image). In vivo data acquired in sheep liver during MR-guided high intensity focused ultrasound (MRgHIFU) sonication were post-processed as proof of applicability in a therapeutic scenario.Bland and Altman mean absolute difference between the novel method and the "ground truth" thermometry in ex vivo static tissue ranged between 0.069 and 0.968, compared to the inherent "white" noise SD of 0.23. The accuracy and precision of the novel method in volunteer liver were found to be on average 0.13 and respectively 0.65 while the inherent "white" noise SD was on average 0.51. The method was successfully applied to large ROIs, up to 6.2 cm inner diameter, and the computing time per slice was systematically less than 100 ms using C++. The current limitations of reference-free PRFS thermometry originate mainly from the need to provide a nearly-closed border, where the MR phase is artifact-free and the tissue is unheated, plus the potential need to reposition that border during breathing to track the motion of the anatomic zone being monitored.A reference-free PRFS thermometry method based on the theoretical framework of harmonic functions is described and evaluated here. The computing time is compatible with online monitoring during local thermotherapy. The current reference-free MRT approach expands the workflow flexibility, eliminates the need for respiratory triggers, enables higher temporal resolution, and is insensitive to unique-event motion of tissue.