-
Notifications
You must be signed in to change notification settings - Fork 7
Description
# MELD Version 2.2 native installation
# OS Ubuntu 22.04
Hello everybody,
I'm experiencing an issue with harmonization in my dataset and would like to ask for help in finding a solution. After preprocessing with FreeSurfer, I initially ran the pipeline with the --harmo_only
option on about 40 subjects, and then ran it again without --harmo_only
for segmentation after preprocessing all subjects. However, I'm encountering the following error (complete log below) for all my subjects:
[neuroCombatFromTraining] In development ...
ERROR: There was an error in the harmonisation of ['sub-FCD000420241031']
It seems that the dict_combat['data']
from the neuroCombatFromTraining
function returns an array with NAs. I'm unsure why this is happening. My dataset consists of about 40 3 Tesla images (T1w + FLAIR), which appear to be of good quality.
My MELD invocation:
./meldgraph.sh new_pt_pipeline.py -id sub-FCD000420241031 -demos ../../derivatives/meld_data/demo_sidses2.csv --harmo_code H250
The subject ID is present in the demographic file and all the subjects are correctly segmented without harmonization.
Thanks in advance for any assistance.
Log:
./meldgraph.sh new_pt_pipeline.py -id sub-FCD0
00420241031 -demos ../../derivatives/meld_data/demo_sidses2.csv --harmo_code H250
Setting MELD_DATA_PATH to /home/orco/data/FCD/derivatives/meld_data
Setting BASE_PATH to /home/orco/data/FCD/derivatives/meld_data/output/preprocessed_surf_data
Setting EXPERIMENT_PATH to /home/orco/data/FCD/derivatives/meld_data/models
Setting FS_SUBJECTS_PATH to /home/orco/data/FCD/derivatives/meld_data/output/fs_outputs
Setting BASE_PATH to /home/orco/data/FCD/derivatives/meld_data/meld_params
setting license/home/orco/data/FCD/code/meld_graph_fork/license.txt
Namespace(id='sub-FCD000420241031', list_ids=None, harmo_code='H250', fastsurfer=False, parallelise=False,
demographic_file='../../derivatives/meld_data/demo_sidses2.csv', harmo_only=False, skip_feature_extraction=
False, no_nifti=False, no_report=False, debug_mode=False)
SCRIPT 1: Call script segmentation
FCD000420241031
BIDS Layout: ...CD/derivatives/meld_data/input | Subjects: 115 | Sessions: 0 | Runs: 0
INFO - sub-FCD000420241031: T1 file used : /home/orco/data/FCD/derivatives/meld_data/input/sub-FCD000420241
031/anat/sub-FCD000420241031_T1w.nii.gz
INFO - sub-FCD000420241031: FLAIR file used : /home/orco/data/FCD/derivatives/meld_data/input/sub-FCD000420
241031/anat/sub-FCD000420241031_FLAIR.nii.gz
STEP 1 - sub-FCD000420241031: Freesurfer outputs already exists for subject sub-FCD000420241031. Freesurfer
will be skipped
STEP 2 - sub-FCD000420241031: Extract surface-based features
INFO - sub-FCD000420241031: Creating registration to template surface
INFO - sub-FCD000420241031: Sampling features in native space
INFO - sub-FCD000420241031: Sample FLAIR features : []
INFO - sub-FCD000420241031: Sample FLAIR features : []
INFO - sub-FCD000420241031: Calculate curvatures
INFO - sub-FCD000420241031: Compute gaussian curvature
INFO - sub-FCD000420241031: Get thickness and white-grey matter contrast
INFO - sub-FCD000420241031: Calculate curvatures
INFO - sub-FCD000420241031: Compute gaussian curvature
INFO - sub-FCD000420241031: Get thickness and white-grey matter contrast
INFO - sub-FCD000420241031: Moving features to template surface
INFO - sub-FCD000420241031: Move feature to xhemi flip
INFO - sub-FCD000420241031: Moving lesion masks to template surface
INFO - sub-FCD000420241031: Creating final training data matrix
scanner for subject sub-FCD000420241031is set as default XT
INFO - sub-FCD000420241031: All features have been extracted and saved in /home/orco/data/FCD/derivatives/meld_data/output/preprocessed_surf_data/MELD_H250
SCRIPT 2: Call script preprocessing
INFO: Use combat parameters from site
STEP: Smoothing feature .on_lh.curv.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
WARNING - sub-FCD000420241031: 15 extremes vertices
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.gm_FLAIR_0.25.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.gm_FLAIR_0.5.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.gm_FLAIR_0.75.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.gm_FLAIR_0.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.pial.K_filtered.sm20.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : no smoothing for this feature
INFO - all : saving data
STEP: Smoothing feature .on_lh.sulc.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.thickness.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.w-g.pct.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
WARNING - sub-FCD000420241031: 10 extremes vertices
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.wm_FLAIR_0.5.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP: Smoothing feature .on_lh.wm_FLAIR_1.mgh
INFO - all: Clip data to remove very extreme values using clip_params_MELD.json
INFO - all : Smoothing with 6 iterations ...
INFO - all : saving data
STEP - sub-FCD000420241031: Regress thickness with curvature
Add feature .on_lh.thickness_regression.sm3.mgh in features
STEP - sub-FCD000420241031: Combat harmonise subjects
STEP: Combat feature .on_lh.thickness.sm3.mgh
[neuroCombatFromTraining] In development ...
ERROR: There was an error in the harmonisation of ['sub-FCD000420241031']