Realtime high resolution 3D multispectral photoacoustic imaging

我研究计划的长期目标是开发快照3D光声成像，用于在高度混浊的介质(如人体组织)中对生化和生物过程进行高分辨率、多光谱、超快成像。快照3D光声成像有可能解决正常组织和病变组织(如癌症)之间的差异，而不需要漫长的扫描程序，也不会对运动伪影敏感。快照3D光声成像代表了一种混合技术，其中单个激光脉冲(5-10 ns)漫射到介质上。作为响应，介质中的光吸收体(目标)产生超声波频率(0.2-10 MHz)的声波。利用连接到数据采集硬件的固定超声换能器阵列并行地检测和记录声波，从而产生成像结果。然后，将光声记录重建成代表目标的位置、大小、形状和光学属性的图像。快照3D光声成像的主要局限性之一是成像质量差，特别是对于生物组织中的光学吸收体等复杂目标。较差的图像质量是由于当前可用的换能器阵列和数据采集硬件的实际通道密度限制的结果。即使装备最好的光声研究小组也只能同时访问不超过256个探测通道。因此，小组已经选择以牺牲快照能力为代价有效地扫描换能器阵列，以获得重建高质量图像所需的较高通道密度。我的团队选择开发用于快照成像的固定阵列，方法是：(I)针对当前可用的通道密度优化换能器阵列设计，以获得与更高密度系统相当的图像质量；(Ii)客观评估优化设计的成像性能；以及(Iii)开发可实际可靠地扩展到数千个通道的新技术。为了实现这三个研究目标，研究计划将建立在我的团队开发的方法的基础上。首先，我们计划使用串扰分析来设计、优化和测试换能器阵列。串扰分析提供成像体积内系统灵敏度的客观测量(独立于图像重建方法)，并且还提供成像体积内换能器信号从一个位置到另一个位置的混叠(重叠)的估计。在整个图像体积中具有高灵敏度和低混叠的换能器阵列设计可以被认为是最优设计。其次，我们计划应用传统的图像质量评估技术，使用信号保真度和感知视觉质量指标来研究换能器阵列设计的成像性能。将为一系列成像任务估计图像质量指标，这些任务为每个优化设计提供了增加几何复杂性的目标。最后，我们将研究将现有数据采集硬件扩展到数千个检测通道的传统方法，并研究一种全新的基于光学的声学换能器的灵敏度和带宽，它有可能被制造成具有数千个独立检测通道的大型阵列。如果我们成功了，那么这项工作的影响将是广泛的，因为它将(I)使世界各地的许多组织能够利用现有的基础设施采用快照3D光声成像技术，(Ii)确定以更低的成本扩展到数千个通道的新检测策略，以及(Iii)使我的团队、合作者和其他人能够为生物医学应用寻求高分辨率超快多光谱3D成像。

The long-term objective of my research program is to develop snapshot 3D photoacoustic imaging for high-resolution, multispectral, ultrafast imaging of biochemical and biological processes in highly turbid media such as tissues of the human body. Snapshot 3D photoacoustic imaging has the potential to resolve differences between normal and diseased tissues (e.g. cancer) without a lengthy scanning procedure and sensitivity to motion artifacts. Snapshot 3D photoacoustic imaging represents a hybrid technology where a single pulse of laser light (5-10 ns) diffusely illuminates a medium. In response, optical absorbers (targets) in the medium generate acoustic waves at ultrasonic frequencies (0.2-10 MHz). Imaging results from detecting and recording, in parallel, the acoustic waves with a stationary array of ultrasonic transducers linked to data acquisition hardware. The photoacoustic recordings are then reconstructed into images representative of the location, size, shape, and optical properties of the targets. One of the main limitations of snapshot 3D photoacoustic imaging is poor image quality, especially for complex targets such as optical absorbers in biological tissue. Poor image quality is a result of practical channel density limitations due of currently available transducer array and data acquisition hardware. Even the best-equipped photoacoustic research groups have access to no more than 256 simultaneous detection channels. Therefore, groups have chosen to either scan the transducer array at the expense of the snapshot capability effectively to obtain the higher channel density needed to reconstruct high quality images. My group has chosen to develop stationary arrays for snapshot imaging by (i) optimizing transducer array designs for currently available channel densities to obtain images of comparable quality to much higher density systems, (ii) objectively evaluating the imaging performance of optimized designs, and (iii) developing new technologies that scale to several thousand channels practically and reliably. To achieve these three research objectives, the research plan will build upon methods developed by my group. First, we plan to design, optimize and test transducer arrays using cross-talk analysis. Cross-talk analysis provides an objective measure of the system sensitivity within the imaged volume (independent of image reconstruction method) and also provides estimates of the aliasing (overlap) of transducer signals from one location to another within the imaged volume. A transducer array design with high sensitivity and low aliasing throughout the image volume can be considered an optimal design. Second, we plan to apply conventional image quality assessment techniques to study imaging performance of the transducer array designs using signal fidelity and perceived visual quality metrics. Image quality metrics will be estimated for a series of imaging tasks that present targets of increasing geometric complexity to each optimized design. Last, we will examine conventional approach to extend existing data acquisition hardware to thousands of detection channels and investigate the sensitivity and bandwidth of a radically new optically-based acoustic transducer, which has potential to be fabricated into a large array with several thousand independent detection channels. If we are successful, then the impact of the work will be widespread as it will (i) enable many groups worldwide to adopt snapshot 3D photoacoustic imaging techniques using existing infrastructure, (ii) identify new detection strategies that scale to several thousand channels with lower cost, and (iii) enable my group, collaborators, and others to pursue high-resolution ultrafast multispectral 3D imaging for biomedical applications.