基于MINFLUX超分辨成像与改进聚类算法的突触纳米结构解析

Nanoscale Synaptic Structure Analysis Using MINFLUX Super-Resolution Imaging and an Improved Clustering Algorithm

摘要: 在细胞尺度的生物成像中，高密度荧光标记导致的信号重叠和结构错误解析限制了纳米级生物复合体的精准建模。本研究结合MINFLUX成像技术，整合现有计算方法并优化参数，提出一种针对突触纳米簇的标准化分析流程。基于现有的三步DBSCAN框架，我们通过参数分级优化与体积筛选策略，提升了密集荧光团的分割精度。利用DCR数据的双峰特性，通过双高斯拟合实现了CaMKII-T286与PSD95标记的高效分色。进一步将多项式回归曲面拟合与AlphaShape表面重构结合，并在MINFLUX数据中实现了突触后膜纳米结构的空间映射。应用该流程分析固定神经元样本，发现PSD95团块与CaMKII-T286寡聚体的高度结合，且PSD95纳米簇呈现网状拓扑。本研究提供了可复用的参数组合方案，为超分辨数据的复杂结构解析提供了参考依据。

Abstract: In cellular-scale biological imaging, high-density fluorescent labeling causes signal overlap and structural misinterpretation, hindering nanoscale biological complex modeling. This study combines MINFLUX imaging with optimized computational methods to develop a standardized analysis workflow for synaptic nanoclusters. We enhance the three-step DBSCAN framework with parameter optimization and volume screening to improve segmentation of dense fluorophores. Using the bimodal property of DCR data and dual Gaussian fitting, we achieve efficient separation of CaMKII-T286 and PSD95 signals. We also combine polynomial regression surface fitting with AlphaShape surface reconstruction for the spatial mapping of postsynaptic membrane nanostructures in MINFLUX data. Applying this workflow to fixed neuronal samples reveals tight association between PSD95 clusters and CaMKII-T286 oligomers, with PSD95 nanoclusters showing a reticular topology. This study offers reusable parameter combinations, providing a reference for analyzing complex structures in super-resolution data.