ddmra.analysis.qcrsfc_summary

qcrsfc_summary(mean_qcs, z_corr_mats, run_covariates=None, alpha=0.05)[source]

Compute descriptive QC-FC benchmark summaries.

These are the standard QC-FC summary statistics reported in the resting-state fMRI denoising literature (e.g., Ciric et al., 2017; Parkes et al., 2018): the median absolute QC-FC correlation and the percentage of edges with a statistically significant QC-FC correlation. Lower values indicate less residual association between run quality and connectivity. These are descriptive diagnostics computed on the raw (non-z-transformed) edge correlations; inferential claims in ddmra are based on the smoothing-curve intercept and slope, not on these summaries.

Parameters:

  • mean_qcs (numpy.ndarray of shape (n_subjects,)) – QC measure (typically mean framewise displacement) across participants.

  • z_corr_mats (numpy.ndarray of shape (n_subjects, n_edges)) – Z-transformed correlation coefficients for ROI-ROI pairs.

  • run_covariates (None or numpy.ndarray of shape (n_subjects, n_covariates), optional) – Run-level covariates to adjust for before correlating QC and FC.

  • alpha (float, optional) – Two-sided significance threshold applied to each edge’s QC-FC correlation. Default is 0.05. The test is uncorrected for multiple comparisons.

Returns:

summary (dict) – Dictionary with the following keys:

  • n_runs: number of runs.

  • n_covariates: number of covariates regressed out.

  • n_edges: number of ROI-ROI edges.

  • median_abs_qcfc: median absolute QC-FC correlation across edges.

  • mean_abs_qcfc: mean absolute QC-FC correlation across edges.

  • n_significant_edges: number of edges with two-sided p < alpha.

  • percent_significant_edges: percentage of edges with two-sided p < alpha.

  • alpha: the significance threshold used.

Notes

Edgewise significance uses the parametric two-sided test for a Pearson correlation, with degrees of freedom reduced by the number of covariates. Under a well-denoised pipeline with no residual QC-FC association, the percentage of significant edges should approach 100 * alpha.