Guide for Beginner#
Pipeline Parameters#
In the nextflow.config file, you can configure various pipeline parameters to customize your processing workflow:
nextflow.config
params {
dataset_dir = "/input" // Input data directory
output_dir = "/output" // Output logs directory
preproc_dir = "${params.output_dir}/preprocessed" // Output results directory
code_dir = "/program/megprep" // All code for preprocessing
do_fs = true // Enable FreeSurfer method
do_only_anatomy = false // if true, only anatomy preprocessing.
is_bids = true // Whether the data is in BIDS format
meg_visualize = true // for coregistration and source_recon
// MRI Parameters
anatomy_preprocess_method = "freesurfer" // "freesurfer" or "deepprep"
anatomy_select_tag = "" //"_run-02_T1w" // defalut:"", <subject_id>+<anatomy_tag> to select multiple runs T1.
// MRI Import datasets
mri_import_config = """
# Filter out specific anatomy, only bids support.
subject_id: '001' # str or null
session_id: null # str or null
task: null
run_id: null
"""
//freesurfer method
t1_dir = "/input" // T1 image directory
t1_input_type = "nifti" // 'dicom' or 'nifti'
//deepprep method
deepprep_device = "cpu"
t1_bids_dir = "/input"
fs_license = "/data/liaopan/megprep/megprep/tools/DeepPrep/license.txt"
//[fixed params]
fs_subjects_dir = "/smri" // FREESURFER SUBJECTS_DIR in Container[fixed]
freesurfer_home = "/opt/freesurfer/7.4.1-1"
// MEG import datasets parameters
dataset_format = 'auto' // Dataset format: 'auto', 'bids', or 'raw'
file_suffix = '.fif' // File suffix for raw datasets
meg_import_config = """
# Filter out specific megs, only bids support.
subject_id: null # str or null
session_id: null # str or null
task: null
run_id: null
#subject_id:
# - '001'
#session_id:
# - '006'
#task:
# - compr
"""
// BEM model parameters
bem_config = """
ico: 4
conductivity:
- 0.3
"""
// MEG Preprocessing Parameters
osl_random_seed = 2025
preproc_config = """
preproc:
#- set_channel_types: {EEG061: eog, EEG062: eog, EEG063: ecg}
- filter: {l_freq: 0.5, h_freq: 40, method: iir, iir_params: {order: 5, ftype: butter}}
- notch_filter: {freqs: 50 100}
- resample: {sfreq: 100}
#- bad_segments: {segment_len: 500, picks: mag, significance_level: 0.1}
#- bad_segments: {segment_len: 500, picks: mag, mode: diff, significance_level: 0.1}
"""
// MEG Artifacts Detection Parameters
artifact_config = """
find_bad_channels:
pyprep:
deviation:
deviation_threshold: 5.0
snr: {}
nan_flat: {}
# hfnoise:
# HF_zscore_threshold: 5.0
# ransac: # very slow
# n_samples: 50
# sample_prop: 0.25
# corr_thresh: 0.75
# frac_bad: 0.4
# corr_window_secs: 5.0
# channel_wise: true
# max_chunk_size: null
# correlation:
# correlation_secs: 1.0
# correlation_threshold: 0.4
# frac_bad: 0.01
psd:
std_multiplier: 6
osl:
ref_meg: auto
significance_level: 0.05
mne:
find_bad_channels_lof:
n_neighbors: 20
picks: mag
metric: euclidean
threshold: 1.5
find_bad_segments:
osl:
segment_len: 1000 # detect_badsegments
mne:
annotate_muscle_zscore:
ch_type: mag
threshold: 12
#annotate_amplitude:
# picks: meg
# annotate_break:
# min_break_duration: 15.0
# t_start_after_previous: 5.0
# t_stop_before_next: 5.0
"""
// MEG ICA Parameters
num_IC = 60 // 0.99999
ICA_random_seed = 2025
ICA_output_dir = "ica_report" // relative path based on preproc dir
// MEG ICA Label Parameters
ic_label_config = """
# detect artifact ICs
ic_ecg: true
ic_eog: true
ic_outlier: true # detect artifact ICs by rules.
find_bads_eog:
ch_name: null # or the ch_name of EOG.
threshold: auto
l_freq: 1
h_freq: 10
start: null
stop: null
measure: zscore
find_bads_ecg:
ch_name: null # or the ch_name of ECG.
threshold: auto
method: ctps
l_freq: 8
h_freq: 16
measure: zscore
find_bads_muscle:
threshold: 0.5
start: null
stop: null
l_freq: 7
h_freq: 45
ICA_classify:
meg_vendor: ctf # neuromag or ctf or quanmag_opm or quspin_opm
explained_var:
threshold: 0.1
ch_type: mag
find_ecg_ics:
time_segment: 10 # seconds
ts_ecg_num_max: 20 # Maximum number of heartbeats expected in the chosen time segment
l_freq: 0.1
h_freq: 10
peak_threshod_coef: 0.4 #Indicates the threshold of the number of ecg signal peak interval (unit: index). (peak_threshod = 0.4 * fs) | # for 1 seconds
peak_std_threshold_coef: 0.05 #Standard deviation threshold of ecg signal peak interval (unit: index). (peak_std_threshold = peak_std_threshold_coef * fs) | # for 1 seconds
find_abnormal_psd_ics:
attention_low_freq: 0
attention_high_freq: 150
le_low_freq: 0
le_high_freq: 12
low_freq_energy_threshold: 0.8 # Threshold above which the component is flagged by low-frequency energy ratio
"""
// MEG Epochs Parameters
epoch_output_dir = "epochs" // relative path based on preproc dir
epoch_config = """
task_type: 'task' # or 'resting'
resting:
fixed_length_duration: 2.0
event_source: 'find_events' # 'event_file' or 'find_events'
autoreject: false # true or false| automatic global_rejection_threshold, get the `reject` params.
#event_file：specific the event type of *_events.tsv | filter | the value of `null` means to get all events.
event_file:
trial_type: null
#type: # you can change `trial_type` to `type` or other type related.
# word_onset_01: 1
# phoneme_onset_01: 2
# trial_type:
# - word_onset_01
# - phoneme_onset_01
# find events
find_events:
#stim_channel: UPPT001 # for CTF Holmes
stim_channel: null
shortest_event: 1
min_duration: 0.0
epochs:
event_id: null
tmin: -0.2
tmax: 1
reject_by_annotation: true
picks: meg
baseline: null
#reject:
#grad: 4000e-13
#mag: 4e-12
preload: true
detrend: null
"""
// MEG-MRI coregistraion Parameters
trans_output_dir = "trans"
core_config = """
omit_head_shape_points: 1 # mm
grow_hair: 0.0 #mm
icp:
n_iterations: 200
lpa_weight: 1.0
nasion_weight: 10.0
rpa_weight: 1.0
hsp_weight: 10.0
eeg_weight: 0.0
hpi_weight: 1.0
finetune_icp:
n_iterations: 200
lpa_weight: 0.0
nasion_weight: 0.0
rpa_weight: 0.0
hsp_weight: 10.0
eeg_weight: 0.0
hpi_weight: 0.0
"""
// Covariance Parameters
covar_output_dir = "covariance"
covar_visualize = true // Whether to generate covariance graphs
covar_type = "epochs" // raw or epochs
raw_covariance_task_id = "resting" // task name
covar_config = """
## 1.Estimate noise covariance matrix from a continuous segment of raw data.
compute_raw_covariance:
tmin: 0
tmax: null
method: auto
reject:
grad: 4000e-13 # T / m (gradiometers)
mag: 4e-12 # T (magnetometers)
reject_by_annotation: true
rank: info
## 2.Estimate noise covariance matrix from epochs.
# find events
events:
stim_channel: null
#stim_channel: UPPT001 # for CTF Holmes
shortest_event: 1
min_duration: 0.0
# For baseline epochs
epochs:
event_id: null # baseline event id
tmin: -0.2 # Start time (in seconds) for covariance calculation window
tmax: 0.0 # End time (in seconds) for covariance calculation window
picks: meg
baseline: null
#reject:
#grad: 4000e-13
#mag: 4e-12
preload: true
detrend: null
reject_by_annotation: true
covariance:
tmin: null #Start time for baseline. If null start at first sample.
tmax: null # End time for baseline. If null end at last sample.
rank: null # Rank used for covariance calculation| meg: 90
"""
// Forward Solution Parameters
fwd_output_dir = "forward_solution"
fwd_epoch_label = "wdonset"
fwd_config = """
surface: white # pial
spacing: ico4
"""
// Source Imaging Parameters
src_output_dir = "source_recon"
src_type = "epochs" // raw or epochs
src_config = """
source_methods:
- dSPM
data_type: meg # mag
spacing: ico4
epoch_label: wdonset
dSPM:
inverse_operator:
loose: auto
depth: 0.8
fixed: auto
rank: info
#meg : 50
apply_inverse:
lambda2: 0.1111111111
method: dSPM
pick_ori: normal
LCMV:
n_rank: 50 # compute_covariance,meg's n_rank
cov_tmin: 0.01
cov_tmax: 0.4
make_lcmv:
reg: 0.05
pick_ori: null
rank:
meg : 50
weight_norm: unit-noise-gain-invariant
“””
Run MEGPrep#
docker run -it --rm \
-v /data/datasets/SMN4Lang:/input \
-v /data/datasets/SMN4Lang/preprocessed:/output \
-v /data/datasets/SMN4Lang/smri:/smri \
-v /data/megprep/license.txt:/fs_license.txt \
-v /data/megprep/nextflow/nextflow.config:/program/nextflow/nextflow.config \
megprep:0.0.3 \
-i /input \
-o /output \
--fs_license_file /license.txt \
--fs_subjects_dir /smri \
--resume