Bíró Péter, H Kovács Bálint Barna, Novák Tibor, Erdélyi Miklós
Department of Optics and Quantum Electronics, University of Szeged, Dóm tér 9, Szeged, 6720, Hungary.
Comput Struct Biotechnol J. 2025 Feb 28;27:920-927. doi: 10.1016/j.csbj.2025.02.037. eCollection 2025.
Single-molecule localization microscopy techniques are one of the most powerful methods in biological studies, allowing the visualization of nanoclusters. Cluster analysis algorithms are used for quantitative evaluation, with DBSCAN being one of the most widely used. Clustering results are extremely sensitive to the initial parameters; thus, several methods including DBSCAN maps, have been developed for parameter optimization. Here, we introduce cluster parameter-based DBSCAN maps, which are directly applicable to measured datasets. These maps can be used for image characterization and parameter optimization through sensitivity studies. We show the applicability of these maps to simulated and measured datasets and compare our results with the recently implemented lacunarity analysis for SMLM measurements.
单分子定位显微镜技术是生物学研究中最强大的方法之一，可实现纳米簇的可视化。聚类分析算法用于定量评估，密度基于空间聚类算法（DBSCAN）是使用最广泛的算法之一。聚类结果对初始参数极为敏感；因此，已经开发了包括DBSCAN映射在内的几种方法用于参数优化。在此，我们介绍基于聚类参数的DBSCAN映射，其可直接应用于测量数据集。这些映射可通过敏感性研究用于图像表征和参数优化。我们展示了这些映射对模拟和测量数据集的适用性，并将我们的结果与最近用于单分子定位显微镜测量的孔隙率分析进行比较。
