-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtechnical_notes.txt
493 lines (307 loc) · 14.1 KB
/
technical_notes.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
+ ============================================================================ +
+ +
+ TECHNICAL NOTES +
+ +
+ ============================================================================ +
+ +
+ This decument is to record the choice of methodology, conducted neuroimaging +
+ data processing, any techinical issues/solutions in Sydney Memory and Ageing +
+ Study (MAS), Older Australian Twins Study (OATS), and Sydney Centenarians +
+ Study (SCS). +
+ +
+ This document is prepared by Dr. Jiyang Jiang ([email protected]) +
+ Dr. Wei Wen ([email protected]) +
+ +
+ ============================================================================ +
1. T1-weighted scans
1.1 Numbers (after removing bad-quality scans)
1.1.1 MAS
N (MW124 raw T1) = 1,227
N (MW1 raw T1) = 541
N (MW2 raw T1) = 424
N (MW4 raw T1) = 262
N (MW12 raw T1) = 146 pairs
N (MW14 raw T1) = 28 pairs
N (MW24 raw T1) = 30 pairs
N (MW124 raw T1) = 201 pairs *
N (MW124 successful cross-sectional recon-all) = 1,220 +
1.1.2 OATS
N (OW123 raw T1) = 948
N (OW1 raw T1) = 421
N (OW2 raw T1) = 290
N (OW3 raw T1) = 237
N (OW12 raw T1) = 75 pairs
N (OW13 raw T1) = 23 pairs
N (OW23 raw T1) = 10 pairs
N (OW123 raw T1) = 200 pairs *
N (OW123 successful cross-sectional recon-all) = 948 +
1.2.3 SCS
* Subjects with all three time points (MW124, OW123) were first picked before looking for two time points (e.g. MW12, OW12).
+ See Section 1.4.3 for not salvageables.
1.2 Scanning parameters
TO-DO : fill in T1w scanning parameters.
1.3 Issue with data
None
1.4 FreeSurfer
FreeSurfer's recon-all was applied to quantify cortical measures, including
cortical thickness, area, volume, etc.
1.4.1 FreeSurfer version
FreeSurfer version 7.1.0 was used. To be specific, freesurfer-linux-centos7_x86_64-7.1.0-20200511-813297b.
1.4.2 recon-all options
Multiple combinations of recon-all options were tested to seek optimal outcome
from recon-all. These include:
recon-all -s <subj> -i <t1_nifti> -all
recon-all -s <subj> -i <t1_nifti> -wsatlas -all #use atlas for skull stripping
recon-all -s <subj> -i <t1_nifti> -bigventricles -all
antsRegistrationSyNQuick.sh -d 3 -f <t1_nifti> -m <flair_nifti> -o <inplaneFLAIR> -t a ; recon-all -subjid <subj> -i <t1_nifti> -FLAIR <inplaneFLAIR> -FLAIRpial -all #use FLAIR to improve pial surface reconstruction
Results showed that
1) skull stripping using atlas (-wsatlas) slightly improved brain extraction results.
2) benefit from adding '-bigventricles' was minimal
3) adding FLAIR messed up pial surface reconstruction. This is likely due to the loose FLAIR acquisition resolution (0.8*0.8*3.5). Ref = https://www.mail-archive.com/[email protected]/msg63099.html. With an isotropic FLAIR acquisition, FLAIR is demonstrated to significantly improve the quality of pial surface reconstruction.
Therefore, MAS, OATS and SCS FreeSurfer recon-all processing was carried out with the following command:
recon-all -s <subj> -i <t1_nifti> -wsatlas -all
1.4.3 recon-all failure not salvageable
1.4.3.1 MAS
Seven MAS scans were not salvageable: 1494_w1,1493_w2,0699_w2,2851_w1,2851_w4,6420_w4,2007_w4
These failures are due to hitting walltime on Gadi cluster which is 48 hours. The reason for the excessive long run is that there are multiple defects to be fixed. FreeSurfer suggested that such defect fixing is taking long time and highly prone to errors.
1.4.3.2 OATS
None
1.4.3.3 SCS
TO DO
1.4.4 Quality control
1.4.4.1 QC procedures
ENIGMA Cortical Quality Control Protocol 2.0 (http://enigma.ini.usc.edu/protocols/imaging-protocols/) was applied for quality control of recon-all results. Briefly,
1) identify subjects with cortical thickness and surface area values that deviate from the rest of your subjects
2) internal surface QC - display slices showing segmentation of cortical ROIs
3) external surface QC - display pial surface on 3D brains.
1.4.4.2 group-level summary statistics
TO BE DONE
1.4.5 Longitudinal pipeline
1.4.6 Hippocampal subfields and nuclei of amygdala
Ref : https://surfer.nmr.mgh.harvard.edu/fswiki/HippocampalSubfieldsAndNucleiOfAmygdala
1.4.6.1 using T1w only
Standard hipppcampual and amygdalar substructure extraction using only T1w.
1.4.6.1.1 cross-sectional
segmentHA_T1.sh <subj> <SUBJECTS_DIR>
1.4.6.1.2 longitudinal
1.4.6.1.3 results extraction to csv
cross-sectional:
longitudinal:
1.4.6.2 using T1w + T2w FLAIR
NOT DONE YET. Also need to compare with T1w-only results.
1.4.6.3 using other MRI modalities
Subfield volumes can also be calculated using other MRI modalities (e.g. DWI or fMRI) using vol2subfield. See Section 6 Multimodal integration at https://surfer.nmr.mgh.harvard.edu/fswiki/HippocampalSubfieldsAndNucleiOfAmygdala.
NOT DONE YET.
1.4.7 Brainstem substructures
Four brainstem substructures : medulla oblongata, pons, midbrain, and superior cerebellar peduncle (SCP). Ref : https://surfer.nmr.mgh.harvard.edu/fswiki/BrainstemSubstructures
1.4.7.1 using T1w only
1.4.7.1.1 cross-sectional
segmentBS.sh <subj> <SUBJECTS_DIR>
1.4.7.1.2 longitudinal
1.4.7.1.3 results extraction to csv
cross-sectional:
longitudinal:
1.4.7.2 using other MRI modalities
https://surfer.nmr.mgh.harvard.edu/fswiki/BrainstemSubstructures (Section 6)
NOT DONE YET.
1.4.8 Thalamic nuclei
parcellation of the thalamus into 25 different nuclei, using a probabilistic atlas built with histological data.
1.4.8.1 using T1w only
1.4.8.1.1 cross-sectional
segmentThalamicNuclei.sh <subj> <SUBJECTS_DIR>
1.4.9 Harmonisation of FreeSurfer results
1.4.9.1 ComBat
1.4.9.1.1 OATS
1.4.9.1.1.1 cross-sectional
1.4.9.1.1.2 longitudinal
1.4.10 Extract measures
1.5 fsl
1.5.1 fsl_anat
fsl_anat with default settings was run on all T1w data
fsl_anat -i <t1_nifti>
1.6 ANTs
1.7 optiBET
optiBET was run on all T1w scans with default parameters for brain extraction:
optiBET.sh -i <t1_nifti>
Ref : https://montilab.psych.ucla.edu/fmri-wiki/optibet/
TO-DO : optiBET processing of some T1w images results in errors. This is to be examined.
1.8 acpcdetect
Default acpcdetect command was run on raw T1w.
acpcdetect -i <t1_nifti>
1.9 c3d
2. Diffusion-weighted imaging
2.1 Numbers
N (MW1 dwi nifti) = 539
N (MW2 dwi nifti) = 423
N (MW4 dwi nifti) = 264
2.2 Scanning parameters
TO-DO : fill in DWI scanning parameters.
2.2 Issue with data
2.2.1 Missing bvec for some MW1 DWI scans
Some MW1 DWI scans missed bvecs. Since MW1 DWI has only 6 gradient directions, they are not likely to be used in studies. Therefore, efforts have not been put to salvage these scans.
2.2.2 Flipped bvec in MW2 DWI scans
The first dimension (along x-axis) of MW2 DWI scans was found to be flipped. The sign of the first dimension needs to be inversed.
TO-DO : check whether MW4 has the same issue.
2.3 MRtrix3 and FSL
2.3.1 mrtrix
MRtrix is used as the main software package to process dwi with FSL's eddy_cuda for preprocessing (see Section 2.3.2). MRtrix processing includes:
a) mrtrix_preprocessing_before-eddy_Gadi_cohort.sh
dwidenoise -force -noise noise.mif in.mif dwidenoise.mif
mrdegibbs -force -axes 0,1 dwidenoise.mif mrdegibbs.mif
mrconvert -force -export_grad_fsl bvec bval mrdegibbs.mif mrdegibbs.nii.gz
dwi2mask -force mrdegibbs.mif mask.mif # dwi2mask was found to be robust in generating masks.
# Ref: https://bookdown.org/u0243256/tbicc/preprocessing-diffusion-images.html#mask-b0
mrconvert -force mask.mif mask.nii.gz
b) fsl_eddy_mas.sh
This runs on GRID. See Section 2.3.2 for details.
2.3.2 fsl
eddy_cuda was applied to dwi data that have gone through MRtrix's dwidenoise and mrdegibbs.
2.3.2.1 eddy_cuda options
eddy_cuda9.1 --imain=mrdegibbs \
--mask=mask \
--acqp=acqparams.txt \
--index=index.txt \
--slm=linear \
--bvecs=bvec \
--bvals=bval \
--repol \
--out=out \
--niter=8 \
--fwhm=10,8,4,2,0,0,0,0 \
--ol_type=sw \
--mporder=8 \
--s2v_niter=8 \
--verbose
Explanations:
--slm linear
MW4 has more than 60 directions, but the vectors do not cover the whole sphere.
--repol
outliers detection and replacement.
--niter=8 --fwhm=10,8,4,2,0,0,0,0
instructs eddy to run 8 initial iterations with decreasing amounts of smoothing. During these iterations eddy estimates eddy current-induced fields, inter-volume movement parameters and detects and replaces outliers. The outlier detection is requested with the --repol flag.
--mporder=8 --s2v_niter=8
tells eddy to run an additional 8 iterations. For the additional iterations the movement model is replaced by a slice-to-vol method with 9 (8+1) degrees of freedom per movement parameter and volume. During these additional iterations the estimates of the eddy current-induced fields and the outliers are also further refined.
--ol_type=sw
Slice-wise (sw) is the default and most basic way of estimating outliers. Since MAS/OATS/SCS DWI data do not use multi-band, only slice-wise outliers are detected.
2.3.2.2 eddy QC
Ref : https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/eddyqc/UsersGuide
NOT DONE YET.
2.4 UKFTractography
3. resting-state BOLD imaging
3.1 Numbers
N (MW2 rsBOLD) = 49
N (MW4 rsBOLD) = 252
N (SCS rsBOLD) = 71
3.2 Scanning parameters
3.3 Issue with data
3.3.1 spatial resolution
3.3.2 inconsistent number of volumes
MW24_cross-sectional has 208 volumes. In OW2_cross-sectional, 31065 has 180 volumes before removing the first 5 volumes. All Queensland scans have 216 volumes. Otherwise, all 208 volumes.
3.4 Preprocessing
3.4.1 fMRIPrep
Standard fMRIprep v1.5.5 commands were conducted with singularity image:
singularity run --cleanenv \
-B ${BIDS_dir}:/data \
-B ${BIDS_dir}/derivatives/fmriprep/${fmriprep_version}:/out \
-B ${BIDS_dir}/derivatives/fmriprep/${fmriprep_version}/work:/work \
/data2/jiyang/mySingulateImgs/fmriprep-${fmriprep_version}.simg \
/data /out \
participant \
--participant_label ${subjID} \
--fs-license-file ${fs_license} \
--work-dir /work \
--ignore {fieldmaps,slicetiming} \
--output-spaces MNI152NLin6Asym:res-2 MNI152NLin2009cAsym \
--skip_bids_validation
In addition to fMRIprep v1.5.5, the following steps were also conducted:
1) nuisance regression (csf, wm, 6 motion params)
2) FWHM = 6 mm
3) bandpass (0.009 - 0.08 Hz)
4) removing the first 5 volumes
Preprocessed fMRI are stored at /data2/public on NiL.
3.4.2 FSL FIX
3.4.2.1 training
Fifty (50) Sydney MAS and 30 SCS scans were randomly selected for training FIX classifier. Subject-level ICs of 30 SCS fMRI scans were manually labelled as signal and noise ICs of different types (TO SPECIFY). Tables below summarized results of leave-one-out cross-validation.
SCS leave-one-out (N = 30)
--------------------------
Thr=1 Thr=2 Thr=5 Thr=10 Thr=20 Thr=30 Thr=40 Thr=50
Mean TPR 97.0 96.5 94.7 92.9 91.1 87.5 84.8 80.8
Mean TNR 58.9 63.9 69.8 76.1 83.9 91.4 95.9 97.7
Mean (3*TPR+TNR)/4 87.5 88.4 88.4 88.7 89.3 88.4 87.5 85.0
Median TPR 100.0 100.0 100.0 100.0 100.0 92.9 88.9 88.2
Median TNR 59.2 65.3 71.9 77.4 86.2 92.7 96.7 97.8
Median (3*TPR+TNR)/4 89.4 90.6 90.8 91.6 93.8 93.0 91.2 90.8
MW24 leave-one-out (N=50)
-------------------------
Thr=1 Thr=2 Thr=5 Thr=10 Thr=20 Thr=30 Thr=40 Thr=50
Mean TPR 98.2 97.8 96.9 96.2 93.0 89.3 85.8 83.3
Mean TNR 63.7 68.9 75.6 80.4 86.0 91.5 94.6 96.5
Mean (3*TPR+TNR)/4 89.6 90.6 91.6 92.3 91.2 89.8 88.0 86.6
Median TPR 100.0 100.0 100.0 100.0 100.0 93.5 92.0 89.4
Median TNR 65.9 68.4 74.6 80.2 85.9 93.1 96.0 97.6
Median (3*TPR+TNR)/4 90.6 91.9 93.0 94.2 93.2 93.8 93.4 90.9
MW24+SCS leave-one-out (N=80)
-----------------------------
Thr=1 Thr=2 Thr=5 Thr=10 Thr=20 Thr=30 Thr=40 Thr=50
Mean TPR 97.6 96.9 96.0 95.1 91.9 87.2 84.9 81.3
Mean TNR 64.6 69.1 74.6 79.4 87.3 92.6 95.0 97.0
Mean (3*TPR+TNR)/4 89.3 89.9 90.6 91.2 90.8 88.6 87.4 85.2
Median TPR 100.0 100.0 100.0 100.0 100.0 92.0 91.3 89.4
Median TNR 66.3 70.3 75.9 81.9 87.8 93.1 96.0 97.9
Median (3*TPR+TNR)/4 91.0 91.8 93.0 94.1 94.9 92.1 92.4 91.7
3.4.2.2 classification
3.4.3 motion outliers
Based on the criterion of translation/rotation along x/y/z axis < 2 mm/degrees (data generated by fMRIPrep v1.5.5), the following subjects should be removed:
MW24_cross-sectional
--------------------
sub-0819A
sub-1128A
sub-1302A
sub-1638A
sub-2046A
sub-2170A *
sub-3001 *
sub-3625A
sub-3761A
sub-3826A
sub-3836A
sub-4612A
sub-4760A
sub-5070A
sub-5163A
sub-5484
sub-5765A
sub-7294
sub-7317A
sub-8233A
sub-8263A
sub-8840A
sub-8925A
sub-8930A
sub-9182A
OW2_cross-sectional
-------------------
sub-11016
sub-11071
sub-12045
sub-12134
sub-400295
sub-524248
SCS
-----------------
sub-100381
sub-100402
sub-200001
sub-200008
* sub-???? = MW2; sub-????A = MW4
3.5 Quality control
3.5.1 MRIQC
3.6 Brain connectivity toolbox
3.7 Dual regression
4. resting-state ASL imaging
4.1 Numbers
4.2 Scanning parameters
4.3 Issue with data
4.3.1 PCASL vs. PASL
4.4 ASLtbx2
4.5 oxford_asl