微血管形态和血流动力学特征与疾病的发生和发展状态密切相关。超声微血管成像是当前超声领域的研究热点，能够对传统超声成像无法显示的百微米级尺寸以下的微血管进行成像。超声微血管成像基于高帧频的平面波发射方式，包括超快速功率多普勒成像（uPDI）和超声定位显微成像（ULM）两种技术。uPDI 基于超声多普勒效应，反映百微米级微血管内的血流量信息。ULM 通过对空间中相互分离的造影剂微泡的中心定位和轨迹追踪，突破衍射极限，能够对十微米级微血管的形态和血流动力学（血流方向、速度）进行成像。然而，平面波成像缺乏发射聚焦，导致发射能量低，进而影响了后续微血管成像的质量（背景噪声高、分辨率低）。本文围绕超声微血管成像这一主题，从方法和应用两个方面分别进行研究。在超声微血管成像方法研究部分：（1）从超声波束合成的角度，利用信号的空间相干性和角度相干性，递进式地提出了空间角度相干因子（SACF）和联合空间角度自适应维纳后滤波（uSA-ASW）两种自适应波束合成方法。仿真和在体实验结果表明，提出的两种波束合成方法显著地提高了有造影和无造影条件下uPDI的对比度和分辨率。进一步，引入深度学习技术，对uSA-ASW 实现了100 倍的计算加速，实现了高质量、快速的自适应波束合成。在体结果表明，该方法在较小的计算负担下，显著提高了ULM 对于大小血管的可视化效果。（2）从超声波束合成后处理的角度，将图像域的非局部均值降噪算法引入到超声射频信号域，利用射频信号包含的丰富相位信息和血流信号在时空域轨迹的连续特征，进行降噪处理。仿体和在体实验结果表明，该方法在有造影和无造影条件下，均能大幅度提高uPDI 的成像质量，实现了对比度和分辨率至少两倍的提升。在超声微血管成像应用研究部分：（1）开展了ULM 用于大鼠糖尿病肾病早期诊断的可行性研究。结果表明，糖尿病大鼠的肾脏微动脉血流速度和血流速度梯度均低于正常组。（2）使用uPDI 评估了不同胎龄早产新生儿的大脑微血管发育情况。结果显示，早产儿大脑微血管密度和直径等参数随胎龄的增加而增加，表明大脑微血管发育的逐渐成熟。综上所述，本文以超声微血管成像的方法和应用为研究内容，从波束合成和波束合成后处理两个角度，提出了在有造影和无造影条件下都适用的高质量超声微血管成像方法，并开展了超声微血管成像的动物实验和临床实验研究。
Studies have shown that microvascular morphology and hemodynamic characteristicsare closely related to the occurrence and development of many diseases. Ultrasoundmicrovascular imaging has become a research hotspot in the current ultrasoundfield, which can visualize the microvessels below the scale of 100 ?m that cannot be detectedby conventional ultrasound imaging modalities. UMI is based on high-frame-rateplane wave imaging, including ultrafast power Doppler imaging (uPDI) and ultrasoundlocalization microscopy (ULM). uPDI is based on ultrasound Doppler effect to reflectthe blood flow volume in microvessels at the scale of 100 ?m. ULM can provide themorphology and hemodynamic (direction and velocity) information of microvessels atthe scale of 10 ?m, breaking throught the diffraction limit by localization and trackingthe spatially-isolated contrast agent microbubbles. Plane wave imaging lacks transmit focusing,resulting in low transmit energy, which in turn affects the quality of subsequentmicrovascular imaging (high background noise and low resolution). This paper focuseson the topic of ultrasound microvascular imaging and conducts research from the perspectivesof methodology and applications.For the studies on the methodology of ultrasound microvascular imaging, (1) fromthe perspective of ultrasound beamforming, two adaptive beamforming methods, spatialangular coherence factor (SACF) and spatial-angular adaptive scaling Wiener postfilter(uSA-ASW), are proposed progressively using the spatial coherence and angular coherenceof signals. Simulations and in vivo experimental results show that the proposedmethods significantly improve the contrast and resolution of uPDI under both contrastenhancedand contrast-free conditions. Furthermore, deep learning (DL) technology is introducedto achieve a 100-fold computational acceleration for uSA-ASW, realizing highqualityand fast adaptive beamforming. In vivo results demonstrate that this DL-basedmethod significantly improves the visualization of vessels with various sizes for ULMunder a relatively low computational burden. (2) From the perspective of post-processingof beamformed signals, the non-local means (NLM) denoising algorithm in the imagedomain is introduced into the ultrasound radio-frequency (RF) domain, where the NLMis performed using the rich phase information contained in the RF signals and the continuoustrajectories of blood flow signals in the spatiotemporal domain. Phantom and in vivo experimental results show that this method significantly improves the imaging qualityof uPDI under both contrast-enhanced and contrast-free conditions, achieving at leasta two-fold increase in contrast and resolution.For the studies on the applications of ultrasound microvascular imaging, preclinicaland clinical studies are conducted. (1) The feasibility of using ULM to early diagnosediabetic kidney disease (DKD) in rats is investigated. The results show that the renal microarterial blood flow velocity and blood flow velocity gradient in DKD rats are lowerthan those in the normal group. (2) uPDI is used to assess the development of cerebralmicrovessels in premature neonates of different gestational ages. The results show thatmicrovascular parameters of premature neonates’ brains, such as microvascular densityand microvascular diameter, increased with gestational age, indicating the gradual maturationof cerebral microvascular development.In summary, this paper focuses on the methodology and applications of ultrasoundmicrovascular imaging. This paper proposes high-quality ultrasound microvascular imagingmethods applicable under both contrast-enhanced and contrast-free conditions, fromthe perspectives of beamforming and post-processing of beamformed signals. In addition,this paper conducts preclinical and clinical studies of ultrasound microvascular imaging.
