Sequeira A M, Ijsselsteijn M E, Rocha M, de Miranda Noel F C C
Department of Informatics, School of Engineering, University of Minho, Braga, Portugal.
Department of Pathology, Leiden University Medical Centre, Leiden, the Netherlands.
Comput Struct Biotechnol J. 2024 Oct 31;23:3920-3928. doi: 10.1016/j.csbj.2024.10.048. eCollection 2024 Dec.
Multiplex spatial proteomic methodologies can provide a unique perspective on the molecular and cellular composition of complex biological systems. Several challenges are associated to the analysis of imaging data, specifically in regard to the normalization of signal-to-noise ratios across images and subtracting background noise. However, there is a lack of user-friendly solutions for denoising multiplex imaging data that can be applied to large datasets. We have developed PENGUIN -Percentile Normalization GUI Image deNoising: a straightforward image preprocessing tool for multiplexed spatial proteomics data. Compared to existing approaches, PENGUIN distinguishes itself by eliminating the need for manual annotation or machine learning models. It effectively preserves signal intensity differences while reducing noise, improving downstream tasks such as cell segmentation and phenotyping. PENGUIN's simplicity, speed, and intuitive interface, available as both a script and a Jupyter notebook, make it easy to adjust image processing parameters, providing a user-friendly experience. We further demonstrate the effectiveness of PENGUIN by comparing it to conventional image processing techniques and solutions tailored for multiplex imaging data.
多重空间蛋白质组学方法能够为复杂生物系统的分子和细胞组成提供独特视角。成像数据分析存在若干挑战，特别是在跨图像的信噪比归一化以及背景噪声扣除方面。然而，目前缺乏适用于大型数据集的、用户友好的多重成像数据去噪解决方案。我们开发了PENGUIN——百分位归一化GUI图像去噪工具：一种用于多重空间蛋白质组学数据的简单图像预处理工具。与现有方法相比，PENGUIN的独特之处在于无需手动注释或机器学习模型。它在有效保留信号强度差异的同时降低噪声，改善诸如细胞分割和表型分析等下游任务。PENGUIN的简单性、速度和直观界面（以脚本和Jupyter笔记本两种形式提供）使得调整图像处理参数变得容易，提供了用户友好的体验。我们通过将PENGUIN与传统图像处理技术以及为多重成像数据量身定制的解决方案进行比较，进一步证明了它的有效性。
