Optoacoustic tomography (OAT) has recently been advanced toward ultrafast volumetric imaging frame rates in the kilohertz range. As a result, excessive data processing and storage capacity requirements are increasingly being imposed on the imaging systems. OAT data commonly exhibit significant sparsity across the spatial, temporal or spectral domains, which facilitated the development of compressed sensing algorithms exploiting various sparse acquisition and under-sampling schemes to reduce data rates. However, performance of compressed sensing critically depends on a priori knowledge on the type of acquired data and/or imaged object, commonly resulting in lack of general applicability and unpredictable image quality. In this work, we report on a fast adaptive OAT data compression framework operating on fully sampled tomographic data. It is based on a wavelet packet transform that maximizes the data compression ratio according to the desired signal energy loss. A dedicated reconstruction method was further developed that efficiently renders images directly from the compressed data. Up to 1000x compression ratios were achieved while providing efficient control over the resulting image quality from arbitrary datasets exhibiting diverse spatial, temporal and spectral characteristics. Our approach enables faster and longer acquisitions and facilitates long-term storage of large OAT datasets.

中文翻译：

---

用于高帧率光声层析成像的通用实时自适应信号压缩

光声断层扫描 (OAT) 最近已朝着千赫兹范围内的超快体积成像帧速率发展。因此，对成像系统的过度数据处理和存储容量要求越来越高。OAT 数据通常在空间、时间或频谱域中表现出显着的稀疏性，这有助于开发利用各种稀疏采集和欠采样方案来降低数据速率的压缩感知算法。然而，压缩感知的性能关键取决于对采集数据和/或成像对象类型的先验知识，通常导致缺乏普遍适用性和不可预测的图像质量。在这项工作中，我们报告了一种在完全采样的断层扫描数据上运行的快速自适应 OAT 数据压缩框架。它基于小波包变换，根据所需的信号能量损失最大化数据压缩比。进一步开发了一种专用的重建方法，可以直接从压缩数据中有效地渲染图像。实现了高达 1000 倍的压缩比，同时提供了对来自具有不同空间、时间和光谱特征的任意数据集的最终图像质量的有效控制。我们的方法可以实现更快、更长时间的采集，并促进大型 OAT 数据集的长期存储。实现了高达 1000 倍的压缩比，同时提供了对来自具有不同空间、时间和光谱特征的任意数据集的最终图像质量的有效控制。我们的方法可以实现更快、更长时间的采集，并促进大型 OAT 数据集的长期存储。实现了高达 1000 倍的压缩比，同时提供了对来自具有不同空间、时间和光谱特征的任意数据集的最终图像质量的有效控制。我们的方法可以实现更快、更长时间的采集，并促进大型 OAT 数据集的长期存储。