sts_cluster.h File Reference

Maris-Oostenveld cluster-mass permutation tests and Threshold-Free Cluster Enhancement for M/EEG inference. More...

#include "sts_global.h"
#include "sts_types.h"
#include <QVector>
#include <QPair>
#include <Eigen/Core>
#include <Eigen/SparseCore>

Include dependency graph for sts_cluster.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	STSLIB::StatsClusterResult
	Per-call output of a cluster permutation test: observed statistic map, cluster masses, cluster p-values and cluster labels. More...
class	STSLIB::StatsCluster
	Maris-Oostenveld cluster-mass permutation tests and Threshold-Free Cluster Enhancement on (channel,time) or (vertex,time) statistic maps. More...

Namespaces
namespace	STSLIB
	Statistical testing (t-tests, F-tests, cluster permutation, multiple comparison correction).

Detailed Description

Maris-Oostenveld cluster-mass permutation tests and Threshold-Free Cluster Enhancement for M/EEG inference.

SPDX-License-Identifier: BSD-3-Clause Copyright (c) 2026 MNE-CPP Authors

Author

Christoph Dinh chris.nosp@m.toph.nosp@m..dinh.nosp@m.@mne.nosp@m.-cpp..nosp@m.org

Since

2.2.0

Date

April 2026

Mass-univariate t- or F-tests on dense (channel, time) or (vertex, time) grids generate a severe multiple-comparison problem. This module implements the standard fix: form an observed statistic map, threshold it at an a-priori cluster-forming level, group the supra-threshold samples into spatially (or spatio-temporally) connected clusters using a sparse adjacency graph from STSLIB::StatsAdjacency and assign each cluster a cluster-mass test statistic equal to the sum of the underlying t- or F-values.

The null distribution of the maximum cluster-mass is then built by Monte-Carlo permutation: condition labels are shuffled for the two-sample test, signs are flipped for the one-sample test, and groups are reassigned for the one-way ANOVA variant. The exchangeable label / sign-flip framework controls family-wise error in the strong sense at the cluster level. The module also exposes Threshold-Free Cluster Enhancement (TFCE), which integrates cluster extent and height over a range of thresholds and removes the arbitrary cluster-forming threshold.

References: Maris & Oostenveld (2007), J. Neurosci. Methods 164(1); Smith & Nichols (2009), NeuroImage 44(1).

Definition in file sts_cluster.h.