Wang Zichen, Hakozaki Hiroyuki, McMahon Gillian, Medina-Carbonero Marta, Schöneberg Johannes
Department of Pharmacology, University of California, San Diego, San Diego, CA, 92093.
Department of Chemistry and Biochemistry, University of California, San Diego, San Diego, CA, 92093.
bioRxiv. 2024 May 30:2024.05.28.596280. doi: 10.1101/2024.05.28.596280.
Light-sheet fluorescence microscopy (LSFM), a prominent fluorescence microscopy technique, offers enhanced temporal resolution for imaging biological samples in four dimensions (4D; x, y, z, time). Some of the most recent implementations, including inverted selective plane illumination microscopy (iSPIM) and lattice light-sheet microscopy (LLSM), rely on a tilting of the sample plane with respect to the light sheet of 30-45 degrees to ease sample preparation. Data from such tilted-sample-plane LSFMs require subsequent deskewing and rotation for proper visualization and analysis. Such transformations currently demand substantial memory allocation. This poses computational challenges, especially with large datasets. The consequence is long processing times compared to data acquisition times, which currently limits the ability for live-viewing the data as it is being captured by the microscope. To enable the fast preprocessing of large light-sheet microscopy datasets without significant hardware demand, we have developed WH-Transform, a novel GPU-accelerated memory-efficient algorithm that integrates deskewing and rotation into a single transformation, significantly reducing memory requirements and reducing the preprocessing run time by at least 10-fold for large image stacks. Benchmarked against conventional methods and existing software, our approach demonstrates linear scalability. Processing large 3D stacks of up to 15 GB is now possible within one minute using a single GPU with 24 GB of memory. Applied to 4D LLSM datasets of human hepatocytes, human lung organoid tissue, and human brain organoid tissue, our method outperforms alternatives, providing rapid, accurate preprocessing within seconds. Importantly, such processing speeds now allow visualization of the raw microscope data stream in real time, significantly improving the usability of LLSM in biology. In summary, this advancement holds transformative potential for light-sheet microscopy, enabling real-time, on-the-fly data processing, visualization, and analysis on standard workstations, thereby revolutionizing biological imaging applications for LLSM, SPIM and similar light microscopes.
光片荧光显微镜（LSFM）是一种卓越的荧光显微镜技术，可为生物样本的四维（4D；x、y、z、时间）成像提供更高的时间分辨率。包括倒置选择性平面照明显微镜（iSPIM）和晶格光片显微镜（LLSM）在内的一些最新技术，依靠将样本平面相对于光片倾斜30 - 45度来简化样本制备。来自此类倾斜样本平面LSFM的数据需要后续的去倾斜和旋转，以进行正确的可视化和分析。目前，此类变换需要大量内存分配。这带来了计算挑战，尤其是对于大型数据集。其结果是与数据采集时间相比，处理时间很长，这目前限制了在显微镜捕获数据时实时查看数据的能力。为了在无需大量硬件需求的情况下实现大型光片显微镜数据集的快速预处理，我们开发了WH - Transform，这是一种新颖的GPU加速且内存高效的算法，它将去倾斜和旋转集成到单个变换中，显著降低内存需求，并将大型图像堆栈的预处理运行时间至少缩短10倍。与传统方法和现有软件相比，我们的方法展示了线性可扩展性。现在，使用具有24GB内存的单个GPU，在一分钟内处理高达15GB的大型3D堆栈成为可能。应用于人类肝细胞、人类肺类器官组织和人类脑类器官组织的4D LLSM数据集时，我们的方法优于其他方法，能在数秒内提供快速、准确的预处理。重要的是，这样的处理速度现在允许实时可视化原始显微镜数据流，显著提高了LLSM在生物学中的可用性。总之，这一进展对光片显微镜具有变革性潜力，能够在标准工作站上进行实时、即时的数据处理、可视化和分析，从而彻底改变LLSM、SPIM和类似光学显微镜的生物成像应用。
