Ahmad Rizwan, Clymer Bradley, Deng Yuanmu, He Guanglong, Vikram Deepti, Kuppusamy Periannan, Zweier Jay L
Department of Electrical and Computer Engineering, The Ohio State University, College of Engineering, Columbus, Ohio 43210, USA.
J Magn Reson. 2006 Apr;179(2):263-72. doi: 10.1016/j.jmr.2005.12.013. Epub 2006 Feb 2.
In electron paramagnetic resonance imaging (EPRI), long data acquisition time is one of the major problems limiting successful and useful biological EPRI experiments. Depending on the configuration (spatial distribution of paramagnetic species), information embedded in some objects can be characterized using a smaller number of projections, while others may require significantly larger number of projections to generate similar results. In order to optimize the acquisition process, it is therefore important to acquire a different number of projections for different objects. In this paper, a prediction scheme is demonstrated that can determine the number of projections required to achieve a preset reconstruction quality for a given object. After acquiring first few projections, corresponding partially filled k-space is analyzed. The complexity of data (to interpolate) in k-space is quantified and used to predict the number of required projections. All the projections are acquired using a mean-square difference-based adaptive acquisition technique that is also demonstrated in this work. The purpose of this non-uniform acquisition is to reduce redundancy in the acquired data which in turn decreases the number of projections required for the given object. It is also demonstrated that the performance of non-uniform acquisition is content dependant, and for certain configurations it may not be as effective as uniform acquisition in preserving signal from low intensity regions. The prediction scheme along with the non-uniform acquisition is tested using computer simulations, imaging of experimental phantoms, and in vivo imaging. Results indicate that the proposed method may save up to 50% of acquisition time. The techniques in this manuscript are described for 2D spatial imaging but can be extended to 3D imaging.
在电子顺磁共振成像（EPRI）中，长数据采集时间是限制成功且有用的生物EPRI实验的主要问题之一。根据配置（顺磁物质的空间分布），某些物体中嵌入的信息可以通过较少数量的投影来表征，而其他物体可能需要显著更多数量的投影才能产生类似的结果。因此，为了优化采集过程，针对不同物体采集不同数量的投影非常重要。在本文中，展示了一种预测方案，该方案可以确定为给定物体实现预设重建质量所需的投影数量。在采集前几个投影后，分析相应的部分填充k空间。对k空间中数据（用于插值）的复杂度进行量化，并用于预测所需的投影数量。所有投影均使用基于均方误差的自适应采集技术进行采集，本文也展示了该技术。这种非均匀采集的目的是减少采集数据中的冗余，进而减少给定物体所需的投影数量。还表明，非均匀采集的性能取决于内容，对于某些配置，在保留低强度区域信号方面可能不如均匀采集有效。使用计算机模拟、实验体模成像和体内成像对预测方案以及非均匀采集进行了测试。结果表明，所提出的方法可节省高达50%的采集时间。本文手稿中的技术是针对二维空间成像描述的，但可扩展到三维成像。
