New Topological Approach to EEG Microstate Analysis Improves Neural State Definition

EEG microstate analysis and related fMRI methods such as hidden Markov models and co-activation pattern analysis are widely used tools for breaking down continuous neural dynamics into a small number of quasi-stable brain states. A persistent limitation of these methods is that nearly all downstream measures depend heavily on template maps defined at the start of analysis, which are derived from clustering voltage maps at global field power peaks. This makes results sensitive to preprocessing steps, clustering algorithms, initialization conditions, and the chosen number of clusters, undermining reproducibility and cross-study comparisons. The authors of this bioRxiv preprint propose treating microstate templates not as cluster centroids but as landmarks embedded in the global geometric structure of a state space built from mutual similarities among scalp voltage maps. A key feature of this reformulation is that it preserves polarity as a meaningful geometric relationship rather than discarding it as analytical redundancy, as conventional pipelines do. The authors report that landmark-based state definitions outperform conventional templates in capturing state structure and improving analytical performance. They argue the core problem in microstate analysis is not merely clustering optimization but how to define valid nodes for coarse-graining continuous dynamics without losing the topology that organizes them, and suggest the framework could also be extended to fMRI.