Freesurfer Group-Level Analysis (mri glmfit) - Revision history

Chris: /* Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based) */

2022-07-21T18:58:24Z

Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based)

← Older revision		Revision as of 14:58, 21 July 2022
Line 77:		Line 77:
	e.g. Monte Carlo correction for surface space (10K iterations, voxel-wise p<.005, positive differences only, no correction for multiple spaces):		e.g. Monte Carlo correction for surface space (10K iterations, voxel-wise p<.005, positive differences only, no correction for multiple spaces):
	mri_glmfit-sim --glmdir glm.wls \		mri_glmfit-sim --glmdir glm.wls \
	--sim mc-z 10000 2.3 mc-z.pos~~.th005~~ --sim-sign pos		--sim mc-z 10000 2.3 mc-z.th005.pos --sim-sign pos

	e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences, cluster-wise p<sub>familywise</sub>=.05 is corrected for lh, rh and mni spaces):		e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences, cluster-wise p<sub>familywise</sub>=.05 is corrected for lh, rh and mni spaces):
	mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 ~~perm~~.abs~~.th01~~ \		mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm10k.th01.abs \
	--sim-sign abs --cwp 0.05 --3spaces		--sim-sign abs --cwp 0.05 --3spaces

	e.g. splitting permutation correction for MNI spaces into 4 parallel jobs for speed (each job will do 2500/10k sims); using -log thresh notation in filenames:		e.g. splitting permutation correction for MNI spaces into 4 parallel jobs for speed (each job will do 2500/10k sims); using -log thresh notation in filenames:
	mri_glmfit-sim --glmdir glm.wls --sim perm 10000 3 perm10k.th30 --sim-sign abs --cwp 0.05 --3spaces --bg 4		mri_glmfit-sim --glmdir glm.wls --sim perm 10000 3 perm10k.th30.abs --sim-sign abs --cwp 0.05 --3spaces --bg 4

	Options for sim type (listed fastest to slowest) are perm, mc-z, and mc-full.		Options for sim type (listed fastest to slowest) are perm, mc-z, and mc-full.

Chris: /* Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based) */

2022-07-21T18:53:27Z

Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based)

← Older revision		Revision as of 14:53, 21 July 2022
Line 79:		Line 79:
	--sim mc-z 10000 2.3 mc-z.pos.th005 --sim-sign pos		--sim mc-z 10000 2.3 mc-z.pos.th005 --sim-sign pos

	e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences, p<sub>familywise</sub>=.05 is corrected for 3 spaces):		e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences, cluster-wise p<sub>familywise</sub>=.05 is corrected for lh, rh and mni spaces):
	mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.th01 \		mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.th01 \
	--sim-sign abs --cwp 0.05 --3spaces		--sim-sign abs --cwp 0.05 --3spaces

Chris: /* Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based) */

2022-07-21T18:52:21Z

Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based)

← Older revision		Revision as of 14:52, 21 July 2022
Line 79:		Line 79:
	--sim mc-z 10000 2.3 mc-z.pos.th005 --sim-sign pos		--sim mc-z 10000 2.3 mc-z.pos.th005 --sim-sign pos

	e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences, ~~correcting~~ for 3 spaces):		e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences, p<sub>familywise</sub>=.05 is corrected for 3 spaces):
	mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.th01 \		mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.th01 \
	--sim-sign abs --cwp 0.05 --3spaces		--sim-sign abs --cwp 0.05 --3spaces

Chris: /* Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based) */

2022-07-21T18:50:29Z

Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based)

← Older revision		Revision as of 14:50, 21 July 2022
Line 75:		Line 75:
	--sim nulltype nsim threshold csdbase \		--sim nulltype nsim threshold csdbase \
	--sim-sign sign		--sim-sign sign
	e.g. Monte Carlo correction for surface space (10K iterations, voxel-wise p<.005, positive differences only):		e.g. Monte Carlo correction for surface space (10K iterations, voxel-wise p<.005, positive differences only, no correction for multiple spaces):
	mri_glmfit-sim --glmdir glm.wls \		mri_glmfit-sim --glmdir glm.wls \
	--sim mc-z 10000 2.3 mc-z.pos.th005 --sim-sign pos		--sim mc-z 10000 2.3 mc-z.pos.th005 --sim-sign pos

	e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences):		e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences, correcting for 3 spaces):
	mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.th01 \		mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.th01 \
	--sim-sign abs --cwp 0.05 --3spaces		--sim-sign abs --cwp 0.05 --3spaces

Chris: /* Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based) */

2022-07-21T18:49:06Z

Correct for Multiple Comparisons with Permutation Test (MNI305 Space, or Surface-Based)

← Older revision		Revision as of 14:49, 21 July 2022
Line 77:		Line 77:
	e.g. Monte Carlo correction for surface space (10K iterations, voxel-wise p<.005, positive differences only):		e.g. Monte Carlo correction for surface space (10K iterations, voxel-wise p<.005, positive differences only):
	mri_glmfit-sim --glmdir glm.wls \		mri_glmfit-sim --glmdir glm.wls \
	--sim mc-z 10000 2.3 mc-z.pos.~~005~~ --sim-sign pos		--sim mc-z 10000 2.3 mc-z.pos.th005 --sim-sign pos

	e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences):		e.g. permutation correction for MNI space (10K iterations, voxel-wise p<.01, absolute value differences):
	mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.01 \		mri_glmfit-sim --glmdir glm.wls --sim perm 10000 2 perm.abs.th01 \
	--sim-sign abs --cwp 0.05 --3spaces		--sim-sign abs --cwp 0.05 --3spaces

			e.g. splitting permutation correction for MNI spaces into 4 parallel jobs for speed (each job will do 2500/10k sims); using -log thresh notation in filenames:
			mri_glmfit-sim --glmdir glm.wls --sim perm 10000 3 perm10k.th30 --sim-sign abs --cwp 0.05 --3spaces --bg 4

	Options for sim type (listed fastest to slowest) are perm, mc-z, and mc-full.		Options for sim type (listed fastest to slowest) are perm, mc-z, and mc-full.

Chris: /* Correct for Multiple Comparisons (Surface-Based) */

2022-07-21T17:30:14Z

Correct for Multiple Comparisons (Surface-Based)

← Older revision		Revision as of 13:30, 21 July 2022
Line 58:		Line 58:

	===Correct for Multiple Comparisons (Surface-Based)===		===Correct for Multiple Comparisons (Surface-Based)===
	After you have run the ~~first-level~~ voxel-wise analysis, you need to correct the contrast maps for multiple comparisons. This is because the previous step runs the analysis on each of many thousand voxels independently, which mathematically implies that ''alpha'' of those significant voxels are actually Type I errors. Again, you will have to perform this step for each analysis directory. This takes very little time to run for the surface-based analyses, which can use precomputed values.		After you have run the voxel-wise analysis, you need to correct the contrast maps for multiple comparisons. This is because the previous step runs the analysis on each of many thousand voxels independently, which mathematically implies that ''alpha'' of those significant voxels are actually Type I errors. Again, you will have to perform this step for each analysis directory. This takes very little time to run for the surface-based analyses, which can use precomputed values.

	An explanation of the parameters: <code>--glmdir glm.wls</code> refers to the directory you specified when you ran the mri_glmfit command. <code>--cache 3 pos</code> indicates that you want to use the pre-cached simulation with a voxel-wise threshold of 10E-3 (i.e., 0.001) and use positive contrast values. <code>--cwpvalthresh .025</code> indicates you will retain clusters with a size-corrected p-value of ''P''<.025 (see the '''Notes''' below).		An explanation of the parameters: <code>--glmdir glm.wls</code> refers to the directory you specified when you ran the mri_glmfit command. <code>--cache 3 pos</code> indicates that you want to use the pre-cached simulation with a voxel-wise threshold of 10E-3 (i.e., 0.001) and use positive contrast values. <code>--cwpvalthresh .025</code> indicates you will retain clusters with a size-corrected p-value of ''P''<.025 (see the '''Notes''' below).

Chris: /* Concatenate First-Level Analyses(isxconcat-sess) */

2022-07-21T16:20:53Z

Concatenate First-Level Analyses(isxconcat-sess)

← Older revision		Revision as of 12:20, 21 July 2022
Line 30:		Line 30:

	Note that if you have already run isxconcat-sess in the past, you may get a warning because an earlier group-level directory will already exist. You should probably just delete the old directory and rerun isxconcatenate-sess.		Note that if you have already run isxconcat-sess in the past, you may get a warning because an earlier group-level directory will already exist. You should probably just delete the old directory and rerun isxconcatenate-sess.
			===Protip for Looping Through Contrasts===
			Contrasts will have a corresponding .config file in one or more of the analysis directories found in $SUBJECTS_DIR. If you plan to automate this process using a script that loops over analysis folders and contrasts, the following code snippet will create a text file containing the names of the contrasts with .config files:
			cd $SUBJECTS_DIR/<span style='color:red'>my_analysis</span>.lh
			ls -1 \| grep config \| sed 's/\.config//g' > ../list_of_contrasts
			#now $SUBJECTS_DIR has a text file called list_of_contrasts for use in a loop

	==Run mri_glmfit==		==Run mri_glmfit==

Chris: /* One-Sample Group Mean */

2021-06-22T18:44:18Z

One-Sample Group Mean

← Older revision		Revision as of 14:44, 22 June 2021
Line 50:		Line 50:
	#contrast 1 in subcortical MNI space - no need to specify the surface		#contrast 1 in subcortical MNI space - no need to specify the surface
	cd ${SUBJECTS_DIR}/RFX/${ANALYSIS}<span style="color:red">.mni</span>/${CONTRAST1}		cd ${SUBJECTS_DIR}/RFX/${ANALYSIS}<span style="color:red">.mni</span>/${CONTRAST1}
	mri_glmfit --y ces~~.nii.gz cesvar~~.nii.gz --osgm --glmdir glm.wls --nii.gz <span style="color:red">--eres-save</span>		mri_glmfit --y ces.nii.gz --osgm --glmdir glm.wls --nii.gz <span style="color:red">--eres-save</span>

	===Correct for Multiple Comparisons (Surface-Based)===		===Correct for Multiple Comparisons (Surface-Based)===

Chris: /* One-Sample Group Mean */

2021-06-22T18:43:19Z

One-Sample Group Mean

← Older revision		Revision as of 14:43, 22 June 2021
Line 50:		Line 50:
	#contrast 1 in subcortical MNI space - no need to specify the surface		#contrast 1 in subcortical MNI space - no need to specify the surface
	cd ${SUBJECTS_DIR}/RFX/${ANALYSIS}<span style="color:red">.mni</span>/${CONTRAST1}		cd ${SUBJECTS_DIR}/RFX/${ANALYSIS}<span style="color:red">.mni</span>/${CONTRAST1}
	mri_glmfit --y ces.nii.gz ~~--wls~~ cesvar.nii.gz --osgm --glmdir glm.wls --nii.gz <span style="color:red">--eres-save</span>		mri_glmfit --y ces.nii.gz cesvar.nii.gz --osgm --glmdir glm.wls --nii.gz <span style="color:red">--eres-save</span>

	===Correct for Multiple Comparisons (Surface-Based)===		===Correct for Multiple Comparisons (Surface-Based)===

Chris: /* One-Sample Group Mean */

2021-06-22T18:38:19Z

One-Sample Group Mean

← Older revision		Revision as of 14:38, 22 June 2021
Line 50:		Line 50:
	#contrast 1 in subcortical MNI space - no need to specify the surface		#contrast 1 in subcortical MNI space - no need to specify the surface
	cd ${SUBJECTS_DIR}/RFX/${ANALYSIS}<span style="color:red">.mni</span>/${CONTRAST1}		cd ${SUBJECTS_DIR}/RFX/${ANALYSIS}<span style="color:red">.mni</span>/${CONTRAST1}
	mri_glmfit --y ces.nii.gz --wls cesvar.nii.gz --osgm --glmdir glm.wls --nii.gz		mri_glmfit --y ces.nii.gz --wls cesvar.nii.gz --osgm --glmdir glm.wls --nii.gz <span style="color:red">--eres-save</span>

	===Correct for Multiple Comparisons (Surface-Based)===		===Correct for Multiple Comparisons (Surface-Based)===