Compressed sensing in synthetic aperture photoacoustic tomography based on a linear array ultrasound transducer
（I）Synthetic aperture configuration with 1/4 data acquisition channel. Limited view reconstruction results (II) and enhanced resolution simulation results (III). The photoacoustic images reconstructed by (a) BP with center aperture (b) BP with synthetic aperture (c) CS with synthetic aperture (d) CS with 1/4 synthetic aperture. Fig. III (e) is the profile along the center line on Figs. III (a)-(d). Note: Back projection (BP), Compressed sensing (CS).
Photoacoustic tomography (PAT) is a noninvasive and non-ionizing hybrid biomedical imaging modality, which has the unique capability of visualizing optical absorption inside the several centimeters deep biological tissue with high acoustic spatial resolution. It is of great value on clinic as a powerful supplement of traditional ultrasound imaging. However, the single linear array transducer based PAT suffers from the limited view challenge and the resolution is still not high enough in current PAT configuration.
In the recent work, a feasible synthetic aperture PAT based on the compressed sensing reconstruction algorithm proposed by Xiangwei Lin, a Ph. D. student in Associate Prof. Mingjian Sun's lab at Harbin Institute of Technology, has overcome these difficulties without any additional data acquisition hardware. This approach combined the ultrasound spatial compounding method to extend the effective aperture size and the compressed sensing technique to reduce the measurement dataset. Both the simulation and experimental results testified the theoretical model and validated that this approach can efficiently improve the image resolution and address the limited view problem while preserving target information with less number of measurements.
This research relies on the synthetic aperture PAT to achieve the multi-view data acquisition to solve the limited view challenge and sparse sampling in the compressed sensing algorithm to recover the target structure of biomedical tissue with reduced measurements. It could provide a potential solution in the clinical transformation to visualize the structure of blood vessel, human breast more clearly and completely with the clinical B-mode ultrasound imaging system. This work is reported in Chinese Optics Letters Volume 15, No. 10, 2017 (Xiangwei Lin, et al., Compressed sensing in synthetic aperture photoacoustic tomography based on a linear-array ultrasound transducer).
“The presented work has greatly improved the resolution of PAT system and its potential for clinical transformation by multi-view sparse data sampling and the compressed sensing reconstruction algorithm”, said Associate Prof. Mingjian Sun.
In the future study, with the optimized sparse sampling, real time 3D synthetic aperture based PAT for whole-body small animal imaging and clinical feasibility for peripheral vessels and human breast imaging will be further investigated.
压缩感知方法在基于超声线阵的合成孔径光声断层成像中的应用
图片说明：(I) 基于压缩感知稀疏采样的合成孔径光声断层成像概念模型，这一方法可用于解决血管仿体的探测角度受限问题 (II) 和实现系统的空间分辨率增强（III）。(a)-(d) 分别展示了中心阵列反投影重建、合成孔径反投影重建、合成孔径稀疏重建、1/4合成孔径稀疏重建结果。图（III）（e）为对空间分辨率的定量分析。
光声成像技术是近年来日益发展起来的一种非侵入式和非电离式的新型生物医学成像方法，具备高光学对比度和高声学分辨率的成像特点。目前基于线性超声阵列的光声断层成像技术作为传统超声成像的有力补充，在临床医学研究上引起了广泛重视，但存在因探测角度受限而造成的组织结构缺失和空间分辨率较低的问题。
哈尔滨工业大学控制科学与工程专业孙明健副教授课题组的博士生蔺祥伟提出了基于压缩感知方法的合成孔径光声断层成像模式，在不增加数据采集硬件资源的条件下，通过合成孔径和压缩感知技术克服了传统线性超声阵列的上述缺点，并分别完成了相关仿真和仿体验证实验。通过基于空间复合的合成孔径技术，可实现对生物组织的三维多角度空间扫描，弥补了因线性阵列角度受限造成的组织结构缺失问题，同时提高了系统的空间分辨率；通过压缩感知方法，可以利用更少的硬件资源完成光声数据采集，在降低硬件成本的基础上保证组织结构的完整性和系统分辨率。
该项研究的创新之处在于将压缩感知方法运用到基于空间复合的合成孔径成像技术中，可在临床现有超声成像设备的基础上完成多角度的稀疏数据采集和压缩感知重建，能够完整展现大血管、人体乳腺组织结构，对光声断层成像技术在临床转化方面的应用具有一定的推动作用。相关研究成果发表在Chinese Optics Letters 2017年第10期上（Xiangwei Lin, et al., Compressed sensing in synthetic aperture photoacoustic tomography based on a linear-array ultrasound transducer）。
孙明健副教授表示，这项工作通过多角度稀疏数据采集和压缩感知重建方法，大大提高了光声断层成像技术的系统分辨率以及其在临床转化方面的应用潜力。
后续研究工作主要是在合成孔径成像中针对特定组织结构，如小鼠全身、大鼠颈动脉、人体乳腺组织、外周血管等，寻找最优稀疏采样率进行压缩感知重建，为实现稀疏合成孔径光声成像的临床应用做出努力。