超高分辨超声成像关键技术研究

The imaging of tumor microvasculature and the quantification of their blood flow is a promising approach for tumor detection at the early stage . However, the structure of microvasculature cannot be resolved by using the standard clinical ultrasound (US) imaging systems due to the fundamental diffraction limit of US waves. Recently, combined with US contrast agents (e.g., microbubble), the super-resolution ultrasound (SR-US) techniques have been proposed, which break this diffraction limit for ultrasound imaging and can achieve the high resolution (~10 um). So, it is possible to image the microvasculature in tumor by SR-US. However, fast, three-dimensional (3-D) SR-US imaging remains a challenge. To address these problems, in this study, we propose a new SR-US imaging approach and establish a new SR-US imaging system. In detail, firstly, based on the compressive sensing technique and the strategy of super-resolution optical fluctuation imaging (SOFI), two kinds of the SR-US imaging methods is established, which is used to identify highly overlapped microbubbles and improve the temporal resolution of SR-US. Secondly, a new tracking method based on the statistic model is established, which is used to generate blood velocity map at super-resolution by tracking individual bubbles through the vasculature. Finally, the effect of different US imaging modality (e..g, STA, SPW, and B-Mode, etc) on the imaging performance of SR-US is systematically evaluated. Based on the obtained optimally parameters, a new SR-US imaging system is established, which can be used to implement 3-D SR-US imaging at high temporal and spatial resolution. The technique is helpful in for the detection, diagnosis, and treatment of disease (tumor).

肿瘤新生血管的检测对于肿瘤的早期诊治具有重要意义。传统的超声成像技术，不能对这些微细血管进行检测。超高分辨超声成像（SR-US）技术在对其检测时仍存在问题，难以实现超高空间分辨的快速三维成像。因此，本项目拟针对SR-US成像的关键问题展开工作，研究并建立适用于肿瘤新生血管检测的SR-US成像方法及平台。主要研究内容包括：（1）基于压缩感知理论与超分辨光学波动成像（SOFI）两种策略，建立SR-US成像算法，提升超分辨结构（血管网络）成像的时间分辨率，实现快速成像。（2）基于概率统计模型，建立微泡运动追踪算法，实现超分辨超声功能（血流速度）成像。（3）研究不同超声成像模态对SR-US成像性能的影响；并在理论分析指导下，整合上述成像算法，建立SR-US成像平台，实现超快地三维超分辨超声结构与功能成像。项目的完成有助于为肿瘤的早期诊断、鉴别诊断等提供更为先进的手段，具有重要的理论和实际应用意义。

超分辨超声成像（SR-US）可实现对深层组织微细血管的检测，对于重大疾病的早期诊断具有重要作用。但是，现有SR-US技术在对微细血管成像时仍存在问题，难以实现临床诊断环境下的有效成像。针对该问题，本项目主要工作目标是研究并建立新型超分辨超声成像方法与平台，实现超快地超分辨结构（血管网络）与功能（血流速度）成像。通过四年研究，我们主要完成了如下工作：1)构建了基于压缩感知理论的SR-US成像方法，结合平面波采集，实现了快速SR-US成像。2)构建并优化了基于SOFI的SR-US成像方法，先后实现了基于复数域、基于主动概率及基于多通道联合SOFI SR-US成像，进一步改善了快速SR-US成像性能。3)建立并优化了基于深度学习的SR-US成像方法，提高了SR-US成像的准确性、鲁棒性及计算效率。在此基础上，实现了超快SR-US成像，部分解决了组织运动对成像结果的影响；实现了定量SR-US成像，根本解决了现有SR-US技术无法恢复幅值信息这一共性问题。4)建立了基于Kuhn-Munkres的微泡追踪算法，实现了超分辨血流速度、血流方向及扩散成像；进一步，通过检测动脉血流速度变化，实现了对高血压肾硬化的早期诊断。5)构建了SR-US成像平台，实现了超快、三维、超分辨超声结构与功能成像；进一步，开展了SR-US成像设备小型化、便携化相关研究。6)与复旦大学附属中山医院、华山医学开展合作，构建了患者肝腹部、脑部的SR-US成像数据库；并探讨了SR-US应用于临床检测的可行性。7)创新性地将声材料理论与超声成像相结合，建立了基于声超构材料的SR-US成像方法。.本项目已在IEEE Trans. Med. Imaging、IEEE Trans. Biomed. Engineering、IEEE Trans. Ultrason. Ferr. Freq. Control、Adv. Funct. Mater.等国内外主流学术期刊和国际会议上发表论文26篇，其中，SCI检索19篇，EI检索6篇；获上海市科技进步二等奖1项（参与）；受理发明专利9项。研究工作得到了国内外同行的高度评价，单篇引用超50次（谷歌学术）。基于上述研究基础，作为负责人先后获批国家自然科学基金面上项目、军科委项目及上海市科委等项目。