DIFFUSION IMAGING USING MIND SAP-EPI

这个子项目是利用由NIH/NCRR资助的中心拨款提供的资源的许多研究子项目之一。子项目和调查员(PI)可能从另一个NIH来源获得了主要资金，因此可能会出现在其他CRISE条目中。列出的机构是针对中心的，而不一定是针对调查员的机构。简介：即使使用现代MRI扫描仪，EPI失真仍然存在问题，特别是在较高图像分辨率或DWI中。最近，我们推出了一种重新设计的螺旋桨EPI，称为SAP-EPI(它具有比以前的螺旋桨EPI方法更高的伪接收器带宽。对于通过使用例如32(频率)256(相位)叶片分辨率获得的高分辨率256？256 SAP-EPI图像，对非共振的敏感性降低了约3倍。在本摘要中，我们将新的SAP-EPI序列与扩散梯度和两种减少失真的附加手段相结合。它们是a)使用GRAPPA的多镜头和b)使用反向梯度极性方法(RGPM)(2，3)的失真校正，这两者都在重新栅格和傅里叶变换到图像域之前应用于叶片。该方法通过最小化两幅相对失真的幅值图像之间的差异来获得无失真的图像。对于EPI成像，通过改变相位斑点梯度的符号来改变失真方向。最近，Andersson et al.(2)和Skare et al.(3)利用基函数(余弦集或三次B-Spline)和光滑、连续的三维场以及三维运动校正，进一步发展了RGPM。RGPM可以通过使用一对相距180°的半傅立叶叶片引入SAP成像。材料和方法：带GRAPPA的多次激发DW-EPI可以减少失真，同时不会由于激发之间的随机相位(由扩散梯度引起)而引入重影。这是通过估计b=0图像上的Grappa权重，并使用这些权重来重建离散余弦图像中的每个镜头来实现的。然后，将每个图像与幅度平均相结合(4)。与带有导航器的多炮DW-EPI不同，单独重建每个炮是保证成功的W.r.t.鬼魂。在这项工作中，我们将SAP-EPI与多达4次的收购相结合。Grappa重量估计在T2w上，并应用于每个DW刀片中的每个喷丸，以创建4个DW刀片。对于256？256的目标分辨率和32？256的刀片分辨率，与传统的256？256ssEPI相比，我们能够减少约12倍的失真。进一步的改进可以通过使用RGPM的2D变体来校正残余失真，该RGPM使用三次B-样条作为基集来构建畸变场(3)。获得了一对相隔180度旋转的螺旋桨叶片。由于它们是在k空间中以相反的方向采集的，因此在这些叶片图像的重建图像中，由于？B0引起的相位累加的影响是相反的。这一点如图2所示。第二个叶片图像被旋转以匹配第一个，随后是2D变形和运动校正，这导致两个校正的叶片和一个？B0场。当零差重建清除叶片图像的相位时，得到k空间中的全尺寸回波中心叶片。对所有其他叶片对重复图2中的处理步骤。对来自所有叶片的校正k空间数据进行重新加密和2D傅立叶变换，以产生最终图像。采集在1.5T GE Excite系统(GEHC，Waukeha，WI)上进行，梯度为50mt/m，志愿者使用8通道相控阵列头线圈(MRI Devices，密尔沃基，WI)。在24°24视野和32°256(频率相)叶片分辨率下获得8个4 mm轴位切片，重建为256°256。结果：1次激发的EPI、1次激发的SAP-EPI、4次增强的SAP-EPI的T2W和DW图像经失真校正后均能显示。对于几何参考，包括一个胖子-Sat FSE。当比较这些图像时，人们应该忽略SNR，因为扫描时间不相同，而专注于失真和模糊的减少。人们可以欣赏SAP-EPI与单次EPI相比的改善。在SAP-EPI中恢复了堆积区附近的构造，但仍保留了一些失真伪影。当使用4次SAP-EPI和RGPM的当前方法时，这些伪影消失，使其类似于FSE图像。参考文献：1.王凤宁，黄泰，林芳华，庄超，陈乃克，钟慧伟，陈彩云，邝国强。《医学评论》2005；54(5)：1232-1240。2.Andersson JL，Skare S，Ashburner J.NeuroImage 2003；20(2)：870-888。3.斯卡雷·S，安德森·J·L.《医学评论》2005；54(1)：169-181。4.斯凯尔·S，纽伯尔德R，克莱顿D，班默，2005年；路易斯湖。致谢：这项工作得到了美国国立卫生研究院(1R01EB002771)、斯坦福高级磁共振技术中心(P41RR09784)、卢卡斯基金会和橡树基金会的部分支持。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Introduction: Even with modern MRI scanners, EPI distortions remain problematic, especially at higher image resolutions or in DWI. Recently, we introduced a redesigned propeller EPI, known as SAP-EPI (that has higher pseudo receiver bandwidth than previous propeller EPI methods. For high resolution 256¿256 SAP-EPI images, obtained by using e.g. 32 (freq.) ¿ 256 (phase) blade resolutions, the sensitivity to off-resonances is reduced by about a factor of 3. In this abstract we combine the new SAP-EPI sequence with diffusion gradients and two additional means of reducing the distortions. These are a) the use of multi-shot with GRAPPA and b) distortion correction using the reversed gradient polarity method (RGPM) (2,3), both of which are applied on the blades prior to regridding and Fourier transformation to the image domain. By this method an undistorted image is achieved by minimizing the difference between two oppositely distorted magnitude images. For EPI imaging, the distortion direction is altered by changing the sign of the phase blip gradients. Recently, Andersson et al. (2) and Skare et al. (3) have developed RGPM further by the use of basis functions (cosine set or cubic 3D B-splines) and smooth and continuous 3D fields as well as 3D motion correction. RGPM may be introduced into SAP imaging by using a pair of half-Fourier blades, oriented 180¿ apart. Materials and methods: Multi-shot DW-EPI with GRAPPA allows reduced distortions while not introducing ghosting due to the random phase between the shots (induced by the diffusion gradients). This is accomplished by estimating GRAPPA weights on the b=0 image and using these weights to reconstruct each of the shots in the DWI¿s. Each image is then combined with magnitude averaging (4). Unlike multi-shot DW-EPI with navigators, reconstructing each shot separately is guaranteed to be successful w.r.t. ghosting. In this work we have combined SAP-EPI with up to 4-shot acquisitions. GRAPPA weights are estimated on the T2w and applied on each shot in each DW blade to create 4 DW blades. For a target resolution of 256¿256 and blade resolution of 32¿256, we are able to reduce distortions by ~12x compared to conventional 256¿256 ssEPI. Further improvement is possible by correcting the residual distortions using a 2D variant of RGPM using cubic B-splines as basis set to build the distortion field (3). A pair of propeller blades rotated 180 degrees apart is acquired. As these are acquired in opposite directions in k-space, the effect of the phase accrual due to ?B0 is reversed in the reconstructed image for these blade images. This is illustrated in Fig. 2. The second blade image is rotated to match the first, followed by the 2D distortion and motion correction, which results in two corrected blades and a ?B0 field. As the homodyne reconstruction clears the phase of the blade images, full-size echo centered blades in k-space result. The processing steps in Fig. 2 are repeated for all other blade pairs. The corrected k-space data from all blades are regridded and 2D Fourier transformed to produce the final image. Acquisition was performed on a 1.5T GE Excite system (GEHC, Waukesha, WI) with 50mT/m gradients on a volunteer using an 8 channel phased-array head coil (MRI Devices, Milwaukee, WI). Eight axial 4 mm slices were acquired with a 24¿24 FOV and 32¿256 (freq.¿phase) blade resolution, reconstructed to 256¿256. Results: T2w and DW images from 1-shot EPI, 1-shot SAP-EPI, 4-shot GRAPPA SAP-EPI with distortion correction are presented. For geometric reference, a fat-sat FSE is included. When comparing these images one should disregard SNR as scan times are not identical, and focus on the distortion and blurring reductions. One can appreciate the improvement with SAP-EPI vs. single-shot EPI. Structures in near the pile-up regions are recovered in SAP-EPI, yet some distortion artifacts remain. When using the current method of a 4-shot SAP-EPI with RGPM, these artifacts disappear, making it similar to the FSE image. References: 1. Wang FN, Huang TY, Lin FH, Chuang TC, Chen NK, Chung HW, Chen CY, Kwong KK. Magn Reson Med 2005;54(5):1232-1240. 2. Andersson JL, Skare S, Ashburner J. Neuroimage 2003;20(2):870-888. 3. Skare S, Andersson JL. Magn Reson Med 2005;54(1):169-181. 4. Skare S, Newbould R, Clayton D, Bammer R. 2005; Lake Louise. p 16. Acknowledgements: This work was supported in part by the NIH (1R01EB002771), the Center of Advanced MR Technology at Stanford (P41RR09784), Lucas Foundation, and Oak Foundation.