基于k-q空间同步加速采集的超高场扩散峰度成像新技术及应用

As a novel magnetic resonance imaging (MRI) technique, diffusion kurtosis imaging (DKI) has demonstrated to accurately quantify molecular diffusion differences in biological tissues, without the injection of contrast agents. This makes DKI unique in detecting microstructure and local pathological mechanism of tissues. However, DKI data acquisition is time-consuming and the spatial resolution of the image is comparably low, which largely limits its clinical application. In recent years, the continuous development of ultra-high field MRI (≥7 T) has greatly reduced time costs and improved spatial resolution of an image. Therefore, the combination of DKI and ultra-high field MRI would be of great significance for fast acquisition of microbiological characteristics and pathological mechanism. To this end, this project aims to establish a novel DKI technique at an ultra-high field MRI scanner. It accelerates the diffusion editing along q space and shortens the image editing along k space in parallel. By combining the two-dimensional principal component analysis-based pattern recognition algorithm, the efficiency of this proposed method is further improved, which offers a practical solution for its bottleneck in the clinical application. Subsequently, by systematically evaluating the feasibility of this proposed technique, it is expected to map the cortex surface and to determine the optimal diffusion metrics for segmenting brain areas and as new evidence for diagnosing neurological diseases in the future. The proposed method is clinically oriented and provides a creative information detection method in medicine, which will benefit a wider context in the medical application.

作为一种既无需注射造影剂、又能准确量化生物组织内分子扩散差异的磁共振方法，扩散峰度成像技术在探测组织微观结构和局部病变机理具有独特优势。目前该技术的数据采集时间较长、影像空间分辨率不高，临床应用极为受限。近年来，不断发展的超高场磁共振（≥7T）在影像数据采集时长和空间分辨率上均有极大改善。将扩散峰度成像技术与超高场磁共振相结合，对快速获取生物微观特征与疾病成因内在机理意义重大。基于此，本项目从扩散磁共振响应方程出发，开发一种q空间加速扩散编辑和k空间缩短影像编辑的新扩散峰度影像技术，并结合二维主成分分析与模式识别算法，大幅度提升工作时效，解决其临床应用受限的瓶颈问题。经系统性评估该技术后，将其应用到人脑组织和皮质分区辨别中，确定人脑最优扩散影像学标记，为脑科学研究和神经系统疾病诊断提供新依据。本项目以临床需求为导向，是医学成像检测领域的新技术、新方法，具有广泛医学应用前景。

扩散峰度磁共振成像（diffusion kurtosis imaging，DKI）利用天然的水分子作为生物示踪剂，是目前可无创对组织微观结构进行探测与成像的唯一方法，在脑梗、肿瘤等疾病的诊断有很大的应用价值。目前该技术的数据采集时间较长、影像空间分辨率不高，临床应用极为受限。本项目利用超高场磁共振在影像数据高信噪比的独特优势，在场强为7特斯拉的磁共振扫描仪上实现并优化扩散峰度成像数据采集方式和图像处理算法，大幅度提升工作时效和图像分辨率，解决其临床应用受限的瓶颈问题。项目首先从扩散磁共振响应方程出发，设计并实现了联合k空间和q空间同步加速的DKI新技术，利用遗传算法优化快速成像的参数，缩短了80%的采集时间；进一步地，利用已开发的二维主成分分析与模式识别算法，对扩散图谱进行高精度重构，与传统方法相比，本项目提出的联合重构算法的峰值信噪比PSNR提高了56%，结构相似度指数SSIM上升了20%。系统地评估该技术后，将其技术移植至目前临床适用的传统扩散张量成像上，利用自动纤维量化技术，用于脑血管病语言损伤和重塑机制研究。本项目以临床需求为导向，对快速获取生物微观特征与疾病成因内在机理意义重大。本项目研发的快速DKI序列在国内外多家医院和科研机构得到了装机使用，并开展了卓有成效的国际学术合作，有力推动了我国在磁共振成像技术领域的进步，也推动了DKI在临床医学诊断和预后评估方面的应用。