基于阵列探测方式的时域光声成像系统

将阵列超声探头和超声相控技术与光声成像相结合的成像系统,与采用水听器的单探头旋转扫描光声成像系统相比,避免了机械旋转机构给光声信号采集所带来的不稳定性,提高了数据采集速度．时域光声信号由64阵元线阵超声探头以电子相控聚焦的方式进行线性扫描采集,然后通过时域后向投影算法进行光声图像的重建．采用波长532nm、重复频率10Hz的脉冲激光,系统可快速重建样品内部光学吸收分部的二维图像,单帧图像数据采集时间小于200s,成像横向分辨率小于2mm．实验结果表明,采用此方法可显著提高系统对光声信号的扫描稳定性和成像效率,该系统是一种有潜在临床应用价值的光声成像系统．     

光声显微镜技术的研究进展

光声显微镜技术具有新兴的一种非侵害性的显微成像技术,具有高分辨率、高对比度、穿透深度高的优点。简要介绍光声成像技术机理,总结报道了国内外几种典型的光声显微成像方法和光声显微图像重建算法的发展历程及其最新进展,指出该技术是一种很有应用前景的医学检测方法。     

合成孔径声成像

本文介绍合成孔径声呐的发展概况,比较多种合成孔径声成像算法。用多阵元方案解决降空间采样率问题,提高测绘速率,建立时域多阵元成像算法,并进行计算机仿真。     

扫描电子声显微镜的成像机理及其应用

扫描电子声显微镜是近几年国际上发展起来的一种新型的显微微观察与成像工具，由于它的成像机理与以往的扫描电镜完全不同，特别是它能进行非破坏性的揭示样品亚表面特性的能力。因此，正越来越多的受到人们的重视与研究。本文主要介绍了扫描电子声显微镜的发展历史，简述了电子声成像技术的基本工作原理。     

基于机器学习的声表面波温度传感器快速优化设计

目的 为提高声表面波谐振器(SAWR)性能,制造高性能声表面波(SAW)温度传感器。方法 通过FEM/BEM理论,建立SAW温度传感器精确仿真优化模型,基于此模型对敏感基片的欧拉角进行大步长优化;同时,结合仿真数据并利用多项式回归模型对敏感基片的欧拉角进行小步长快速优化。结果 文中提出的FEM/BEM仿真模型与机器学习相结合优化设计方法不仅能够实现SAWR的精确模拟,而且可大幅提高优化效率。优化结果与实际器件的中心频率相对误差为0.4%,Q值相对误差为1.2%。文中提出的FEM/BEM仿真模型与机器学习相结合优化设计方法与纯FEM/BEM方法相比,单个切型计算速度提高了2 000多倍。结论 所设计的优化系统可用于谐振器敏感基片切型的快速优化设计,可缩短高性能SAW温度传感器的开发周期。     

基于光声差分衰减频谱的骨质评估仿真研究

光声层析成像术(Photoacoustic Tomography,PAT)在骨组织微结构的量化评估方面具有潜力,但在传统PAT工作模式下,松质骨的固液两相多孔结构导致骨小梁等分布式光吸收成分激发的光声信号混叠,增加了定量分析骨微结构特性的难度和复杂度。针对这一问题,文章将PAT系统改进为偏心激励-差分检测模式,获取差分衰减频谱(Differential Attenuation Spectrum,DAS);并通过数值仿真计算和验证了松质骨孔隙率与光声差分衰减频谱特征参数的相关性。研究结果表明：提取的光声差分衰减频谱特征参数与骨头孔隙率呈强线性相关,基于光声差分衰减频谱的分析方法可有效实现骨质定量评估和诊断。     

应用光声技术测定薄型材料导温系数的研究

本文报导了用光声技术测试和研究薄型试样导温系数的结果,建立了该方法的理论模型,介绍了实验装置和数据处理的方法。用本装置对三种标准和亚标准试样的测试结果表明,实测值和文献推荐值相当吻合。本文的基本理论和实验方法同样也适用于其它有机、无机等薄型材料。     

固体光声检测的两维理论

本文给出了四层媒质理论模型中有两层具有光吸收特性时，各层媒质中两维交变温度场的严格解，并由此导出目前光声检测中常用的传声器检测，光热光偏转检测，热透镜、光热光位移检测、光声压电和热释电检测中先声信号的理论表达式，使固体光学检测理论更系统化和实用化。     

基于无监督学习的高分辨光声重建

光声断层成像(Optoacoustic Tomography,OAT)是一种新兴的生物医学成像技术,在基础医学研究与临床实践中具有重要作用。针对现有光声断层成像空间分辨率较低的问题,提出了一种结合物理点扩散函数(Point Spread Function,PSF)模型和卷积神经网络(Convolutional Neural Network,CNN)的新型高分辨光声重建网络方法(Physical Attention U-Net,Phys-AU-Net)。该方法采用无监督学习策略,结合物理PSF模型和基于注意力机制的U-Net网络。其中,物理PSF模型用于完成对衍射受限机制的模拟,基于注意力机制的U-Net网络用于实现对高密度重叠吸收体图像的特征提取。在二者共同作用下,Phys-AU-Net突破了声衍射极限对于OAT成像空间分辨率的限制。实验结果表明,Phys-AU-Net能够有效实现对声衍射受限光声断层图像的高分辨重建,其性能相较于U-Net网络具有较大程度提升,在结构相似性指标(Structural Similarity,SSIM)方面提升了43.5%,在峰值信噪比(Peak Sign...     

一种新的超声成像方法振动声成像

文章介绍一种新的超声成像方法———振动声成像。首先从理论上分析了振动声成像的基本原理;然后分别介绍了该方法在钙化动脉血管成像、乳房中微钙化灶的检测以及受热疗和HIFU治疗的肿瘤组织焦斑检测等方面的应用;最后比较了振动声成像与传统超声成像不同点并指出了振动声成像独特的优势。     

机器学习分布式网络传输异常数据智能检测方法

对于链路状态数据库的网络传输异常数据检测存在检测数据不完整、较为敏感、检测效率差的问题,提出基于机器学习的分布式网络传输异常数据智能检测方法,通过K最近邻分簇算法对分布式网络节点实施分簇,利用贝叶斯分类算法检测簇头是否出现异常;确定异常簇后,选取小波阈值降噪方法对异常簇内数据进行降噪处理,在此基础上,采用遗传算法检测降...     

基于机器学习的轨道角动量光束模式探测技术研究进展

轨道角动量(OAM)复用和编码技术可有效提高光通信系统信道容量。近些年研究者提出将机器学习(ML)技术用于OAM模式探测以提高OAM光通信系统性能。本文对基于机器学习的OAM模式探测方案进行了综述,包括误差反向传播(BP)神经网络、自组织神经网络(SOM)、支持向量机(SVM)、卷积神经网络(CNN)、光束变换辅助的识别技术以及全光衍射深度神经网络(D2NN),分析了各类机器学习OAM探测器在对抗大气、水下信道带来的干扰时展现出的性能差异以及各自优势。

     

基于机器学习的热导率预测模型研究

传统的理论研究、实验研究及计算仿真已无法满足科学家对新材料的探索与设计。数据驱动的机器学习算法对材料的筛选与性能预测有着推动作用。将机器学习算法应用到材料信息学,基于现有材料热导率数据集,建立机器学习热导率预测模型,通过交叉验证来对机器学习回归模型进行评估。利用机器学习算法建立描述符与热导率属性之间的映射模型,可用于大规模的材料筛选,从而指导实验研究。     

基于深度学习的超声成像技术研究现状

超声成像技术以其无损、无辐射、实时性好、成像深度深和检查费用低等优点,是目前临床医学最常应用的影像技术之一。随着医疗技术的日益发展,兼顾高帧率和高质量的超声成像已成为临床上的迫切需求。深度学习作为能快速提取信号特征的技术,近年来已经在超声成像领域展开多种应用研究并产生了较好的效果,具有很大的应用前景。文章总结了基于深度学习的超声成像技术现状,重点介绍了超声成像技术中常用深度学习架构,列举了深度学习技术在超声成像中的应用,最后总结了目前深度学习在训练以及实现临床应用中可能会遇到的挑战。     

采用稀疏测量数据的有限角度光声层析成像的研究进展

生物光声层析(Photoacoustic Tomography,PAT)成像可以反映生物组织的光吸收分布,定量测量组织的光吸收系数和散射系数,进而分析组织成分,为疾病的早期诊断和治疗提供可靠的依据。由于成像目标特殊的几何结构以及成像装置的机械结构、空间位置和成像时间等的限制,超声探测器只能在有限的角度范围内扫描,采集到稀疏的光声测量数据,导致重建图像中出现伪影和失真。针对有限角度扫描和稀疏测量数据问题,对目前主流的光声图像重建算法进行综述和分析。     

Deep learning and machine learning-based early survival predictions of glioblastoma patients using pre-operative three-dimensional brain magnetic resonance imaging modalities

To propose and implement an automated machine learning (ML) based methodology to predict the overall survival of glioblastoma multiforme (GBM) patients. In the proposed methodology, we used deep learning (DL) based 3D U-shaped Convolutional Neural Network inspired encoder-decoder architecture to segment the brain tumor. Further, feature extraction was performed on these segmented and raw magnetic resonance imaging (MRI) scans using a pre-trained 2D residual neural network. The dimension-reduced principal components were integrated with clinical data and the handcrafted features of tumor subregions to compare the performance of regression-based automated ML techniques. Through the proposed methodology, we achieved the mean squared error (MSE) of 87 067.328, median squared error of 30 915.66, and a SpearmanR correlation of 0.326 for survival prediction (SP) with the validation set of Multimodal Brain Tumor Segmentation 2020 dataset. These results made the MSE far better than the existing automated techniques for the same patients. Automated SP of GBM patients is a crucial topic with its relevance in clinical use. The results proved that DL-based feature extraction using 2D pre-trained networks is better than many heavily trained 3D and 2D prediction models from scratch. The ensembled approach has produced better results than single models. The most crucial feature affecting GBM patients' survival is the patient's age, as per the feature importance plots presented in this work. The most critical MRI modality for SP of GBM patients is the T2 fluid attenuated inversion recovery, as evident from the feature importance plots.     

Theoretical design of vascular imaging based on hall effect

The conceptual technique of vascular imaging and blood flow functional imaging based on Hall Effect is presented in this article. With this non‐invasive approach, both 3D anatomical imaging of the vasculature and functional imaging of blood flow in the deep structure of the human body can be obtained without radiation risk. The technique is based on the fact that the induced charges can be generated when the blood flows through the magnetic field. The induced electric field strength is measured by two groups of detector arrays, which captures not only the position of vasculature in each section, but also the velocity of blood flow and vessel size. The captured images can also be used for 3D reconstruction of the anatomical models. The designed system architecture including both hardware and software is described. © 2013 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 23, 85–96, 2013     

Environment-dependent generation of photoacoustic waves from plasmonic nanoparticles

Nanoparticle-augmented photoacoustics is an emerging technique for molecular imaging. This study investigates the fundamental process of the photoacoustic signal generation by plasmonic nanoparticles suspended in a weakly absorbing fluid. The photoacoustic signal of gold nanospheres with varying silica shell thicknesses is shown to be dominated by the heat transfer between the nanoparticles and the surrounding environment.     

Safety profile of two-dimensional Pd nanosheets for photothermal therapy and photoacoustic imaging

Two-dimensional (2D) nanosheets have emerged as an important class of nanomaterial with great potential in the field of biomedicines,particularly in cancer theranostics.However,owing to the lack of effective methods that synthesize uniform 2D nanomaterials with controlled size,systematic evaluation of size-dependent bio-behaviors of 2D nanomaterials is rarely reported.To the best of our knowledge,we are the first to report a systematic evaluation of the influence of size of 2D nanomaterials on their bio-behaviors.2D Pd nanosheets with diameters ranging from 5 to 80 nm were synthesized and tested in cell and animal models to assess their size-dependent bioapplication,biodistribution,elimination,toxicity,and genomic gene expression profiles.Our results showed size significantly influences the biological behaviors of Pd nanosheets,including their photothermal and photoacoustic effects,pharmacokinetics,and toxicity.Compared to larger-sized Pd nanosheets,smaller-sized Pd nanosheets exhibited more advanced photoacoustic imaging and photothermal effects upon ultralow laser irradiation.Moreover,in vivo results indicated that 5-nm Pd nanosheets escape from the reticuloendothelial system with a longer blood half-life and can be cleared by renal excretion,while Pd nanosheets with larger sizes mainly accumulate in the liver and spleen.The 30-nm Pd nanosheets exhibited the highest tumor accumulation.Although Pd nanosheets did not cause any appreciable toxicity at the cellular level,we observed slight lipid accumulation in the liver and inflammation in the spleen.Genomic gene expression analysis showed that 80-nm Pd nanosheets interacted with more cellular components and affected more biological processes in the liver,as compared to 5-nm Pd nanosheets.We believe this work will provide valuable information and insights into the clinical application of 2D Pd nanosheets as nanomedicines.     

一种黄金分割优化的极限学习机算法

针对极限学习机(ELM)存在大量隐层神经元个数和随机给定权值导致算法性能不稳定等问题,将黄金分割法(Golden Section)与ELM相结合提出了基于黄金分割优化的极限学习机算法(GS-ELM).首先通过黄金分割法对ELM隐含层节点数进行优化,接着再用该方法对ELM输入层权值和隐含层偏差进行优化.实验结果表明,相比较传统的BP神经网络,支持向量机和极限学习机,GS-ELM算法能获得较高的分类精度.