Settings Description (settings.ini)

Settings Description (`settings.ini`)#

MEGqc behavior is controlled through settings.ini. You can use package defaults, a global custom config, or per-dataset config files.

General section#

Variable	Default	Type	Description
`ch_types`	`mag, grad`	string list	Channel types to process.
`STD`	`True`	bool	Enable standard deviation metric.
`PSD`	`True`	bool	Enable power spectral density metric.
`PTP_manual`	`True`	bool	Enable manual peak-to-peak metric.
`PTP_auto_mne`	`False`	bool	Enable MNE auto peak-to-peak metric.
`ECG`	`True`	bool	Enable ECG correlation metric.
`EOG`	`True`	bool	Enable EOG correlation metric.
`Head`	`False`	bool	Enable head-movement metric (requires cHPI/head localization info).
`Muscle`	`True`	bool	Enable muscle-artifact metric.
`data_crop_tmin`	`0`	float	Start crop time in seconds.
`data_crop_tmax`	blank	float/blank	End crop time in seconds; blank uses full recording.

Filtering section#

Variable	Default	Type	Description
`apply_filtering`	`True`	bool	Enable filtering stage.
`downsample_to_hz`	`1000`	float/int	Downsample target frequency.
`l_freq`	`0`	float/int	High-pass edge (Hz).
`h_freq`	`140`	float/int	Low-pass edge (Hz).
`method`	`iir`	string	Filter method.

Epoching section#

Variable	Default	Type	Description
`event_dur`	`0.2`	float	Event duration (s).
`epoch_tmin`	`-0.2`	float	Epoch start relative to event (s).
`epoch_tmax`	`1`	float	Epoch end relative to event (s).
`stim_channel`	blank	string	Optional explicit stim channel.
`event_repeated`	`merge`	string	Duplicate-event handling (`error`, `drop`, `merge`).
`use_fixed_length_epochs`	`True`	bool	Fallback epoching when event structure is unavailable.
`fixed_epoch_duration`	`2.0`	float	Fixed-length epoch size (s).
`fixed_epoch_overlap`	`0.0`	float	Overlap between fixed epochs (s).

Metric sections#

STD (Standard Deviation)#

Variable	Default	Type	Description
`std_lvl`	`1`	int	Number of standard deviations from mean for threshold.
`allow_percent_noisy_flat_epochs`	`70`	int	Max % of epochs a channel can be noisy/flat before marking the channel.
`noisy_channel_multiplier`	`1.2`	float	Multiplier for noisy channel detection (higher = fewer noisy flags).
`flat_multiplier`	`0.5`	float	Multiplier for flat channel detection.

PSD (Power Spectral Density)#

Variable	Default	Type	Description
`freq_min`	`0.5`	float	Lower frequency for PSD calculation (Hz).
`freq_max`	`140`	float	Upper frequency for PSD calculation (Hz).
`psd_step_size`	`1`	float	Frequency resolution of PSD (Hz).

PTP_manual (Peak-to-Peak Manual)#

Variable	Default	Type	Description
`numba_version`	`True`	bool	Use Numba-accelerated calculation for speed.
`max_pair_dist_sec`	`20`	float	Maximum pair distance in seconds.
`ptp_thresh_lvl`	`10`	int	Scaling factor for threshold (higher = more peaks detected).
`allow_percent_noisy_flat_epochs`	`70`	int	Max % of epochs for noisy/flat marking.
`std_lvl`	`1`	int	Standard deviations from mean for threshold.
`noisy_channel_multiplier`	`1.2`	float	Multiplier for noisy channel detection.
`flat_multiplier`	`0.5`	float	Multiplier for flat channel detection.
`ptp_top_limit`	`1e-12`	float	Reserved: absolute upper limit (Tesla).
`ptp_bottom_limit`	`-1e-12`	float	Reserved: absolute lower limit (Tesla).

PTP_auto (Peak-to-Peak Auto - MNE-based)#

Variable	Default	Type	Description
`peak_m`	`4e-14`	float	Min PtP amplitude for magnetometer peaks (T).
`peak_g`	`4e-14`	float	Min PtP amplitude for gradiometer peaks (T/m).
`flat_m`	`3e-14`	float	Max PtP amplitude for magnetometer flat detection (T).
`flat_g`	`3e-14`	float	Max PtP amplitude for gradiometer flat detection (T/m).
`bad_percent`	`5`	int	% of time above threshold to mark as bad.
`min_duration`	`0.002`	float	Min duration (s) above threshold to count.

ECG (Electrocardiogram)#

Variable	Default	Type	Description
`drop_bad_ch`	`True`	bool	Drop bad ECG channel and reconstruct from magnetometers.
`n_breaks_bursts_allowed_per_10min`	`3`	int	Allowed breaks in ECG per 10 minutes.
`allowed_range_of_peaks_stds`	`0.14`	float	Allowed peak height variation (0-1 scaled).
`height_multiplier`	`0.6`	float	Peak height threshold multiplier.
`norm_lvl`	`1`	int	Normalization level for threshold scaling.
`gaussian_sigma`	`4`	int	Gaussian smoothing kernel sigma.
`thresh_lvl_peakfinder`	`5`	int	Peak finder threshold level.
`fixed_channel_names`	blank	string	Explicit ECG channel names (comma-separated).

EOG (Electrooculogram)#

Variable	Default	Type	Description
`n_breaks_bursts_allowed_per_10min`	`3`	int	Allowed breaks in EOG per 10 minutes.
`allowed_range_of_peaks_stds`	`0.15`	float	Allowed peak height variation (0-1 scaled).
`norm_lvl`	`1`	int	Normalization level for threshold scaling.
`gaussian_sigma`	`6`	int	Gaussian smoothing kernel sigma (higher for noisier EOG).
`thresh_lvl_peakfinder`	`3`	int	Peak finder threshold level.
`fixed_channel_names`	blank	string	Explicit EOG channel names (comma-separated).

Muscle#

Variable	Default	Type	Description
`muscle_freqs`	`110, 140`	int pair	Frequency band for muscle detection (Hz).
`threshold_muscle`	`5`	int	Z-score threshold for muscle artifact detection.
`min_length_good`	`0.2`	float	Min duration of “good” data between muscle events (s).
`min_distance_between_different_muscle_events`	`1`	float	Min distance between separate muscle events (s).

Global Quality Index section (`[GlobalQualityIndex]`)#

Variable	Default	Type	Description
`compute_gqi`	`True`	bool	Enable GQI computation.
`include_ecg_eog`	`True`	bool	Include ECG/EOG correlation term in GQI.
`bad_ch_start`	`0`	int	% bad channels below which no penalty applies.
`bad_ch_end`	`100`	int	% bad channels at which maximum penalty applies.
`bad_ch_weight`	`35`	int	Contribution weight for channel-variability penalty.
`correlation_start`	`0`	int	% correlated channels below which no penalty applies.
`correlation_end`	`100`	int	% correlated channels at which maximum penalty applies.
`correlation_weight`	`30`	int	Contribution weight for ECG/EOG-correlation penalty.
`muscle_start`	`0`	float	Muscle ratio below which no penalty applies.
`muscle_end`	`0.0001`	float	Muscle ratio at which maximum penalty applies.
`muscle_weight`	`15`	int	Contribution weight for muscle penalty.
`psd_noise_start`	`0`	int	% PSD noise below which no penalty applies.
`psd_noise_end`	`100`	int	% PSD noise at which maximum penalty applies.
`psd_noise_weight`	`20`	int	Contribution weight for PSD-noise penalty.

Note: Weights are expressed as integers and are automatically normalized so they sum to 100%. See Global Quality Index for detailed interpretation.

Notes#

GUI settings editors prevent many manual formatting errors by using typed controls.
In CLI, keep numeric formatting valid INI syntax (no units in value fields).
Prefer profiles (analysis_mode=new/reuse/latest) when testing multiple settings variants.

For practical examples, see: