Software development for high-throughput analysis of microscopy data

全载玻片扫描（WSS）和双光子活体显微镜（2P-IVM）都是功能强大且用途广泛的技术，在高空间分辨率下分析生物事件非常有价值。然而，WSS生成高分辨率数据的代价是对原始计算能力的要求非常高，因为生成的数据量特别大。此外，由于缺乏解剖区域的自动检测，整个器官切片的高通量分析受到严重损害。2P-IVM特有的另一个挑战来自于来自心脏和呼吸周期以及肌肉张力和蠕动的不良运动伪影。为了正确地解释数据，需要防止或纠正这些运动伪影。因此，深入分析复杂生物结构中的生物事件需要开发高效的软件解决方案来处理大量的显微镜数据。因此，基于我们已发表的工作，在AIM 1中，我们将开发用于高通量分析来自全器官切片WSS的大量图像的软件模块，用于肺，肠和脑分析。我们将开发新的算法，用于小鼠脑切片的神经解剖检测，以量化蛋白质表达或计数特定区域的细胞。我们还将改进算法，优化处理速度。AIM 2将致力于开发一种自动化、交互式图像处理和分析软件，为科学界提供一个交钥匙解决方案，以在当前方法允许的一小部分时间内获得最大限度的成像数据。开发已经开始使用MATLAB环境，利用矢量化，并行计算和GPGPU加速技术来利用其大部分的高计算能力。将进行算法开发、软件编码和测试、用户界面设计、用户需求和功能规格的创建，以及最终用户文档的创建，以便在易用性、多功能性和处理能力方面为最终用户提供增强的体验。这项研究计划将为科学界提供非常有用的工具，将WSS和2P-IVM的范围扩大到所有研究领域的生物医学科学家。

Whole slide scanning (WSS) and two-photon intravital microscopy (2P-IVM) are both powerful and versatile techniques that are extremely valuable to analyze biological events at high spatial resolution. However, the generation of high-resolution data by WSS comes at the price of being exceedingly demanding in term of raw computing power since the amount of data generated is particularly large. Moreover, high-throughput analysis of whole organ sections is strongly impaired by the lack of automated detection of anatomical regions. Another challenge specific to 2P-IVM originates from undesirable movement artifacts coming from cardiac and respiratory cycles as well as muscle tone and peristalsis. These motion artifacts need to be prevented or corrected for proper data interpretation. Therefore, in-depth analysis of biological events in complex biological structures requires the development of efficient software solutions to process large sets of microscopy data. Hence, based on our published work, in AIM 1, we will develop software modules for high-throughput analysis of large sets of images coming from WSS of whole organ sections, for lung, gut and brain analysis. We will notably develop new algorithms for neuroanatomical detection in mouse brain section in order to quantify protein expression or to count cells in specific regions. We will also refine our algorithms and optimize the processing speed. AIM 2 will be devoted to the development of an automated, interactive image processing and analysis software to provide the scientific community with a turnkey solution to get the most out of their imaging data in a fraction of the time current methods would allow them to. Development has already been initiated using MATLAB environment, taking advantage of vectorization, parallel computing and GPGPU acceleration techniques to harness the most of its high computational capabilities. Algorithm development, software coding and testing, design of user interface, creation of user requirements and functional specifications, and creation of end user documentation will be carried out in order to provide end-users with enhanced experience in terms of ease-of-use, versatility and processing power. This research program will deliver extremely useful tools for the scientific community, expanding the reach of WSS and 2P-IVM to biomedical scientists of all research areas.