Hi Doug,
I am following up your reply to this issue as I am still trying to find a solution to this issue. For your questions - yes, I reviewed the images in freeview (and fsleyes) and there are no obvious issues with them (i.e., neither are missing any part of the brain and both are decent T1 images). I also tried using the unmasked (whole head) images, but the same MyMatrix error happens. The --debug output is pasted below and attached as a text file. If it will help, I can share a couple of participant images from our GE scanner.

Interestingly, while including the –affine flag works for some participants, for others I get “MatrixLog Error: Result too imaginary to ignore!” error.

Any help/recommendations are greatly appreciated.

Thanks

Andrew

----
Andrew Beaudin, PhD | Research Associate
University of Calgary | Cumming School of Medicine
Department of Clinical Neurosciences
HSC Rm 2935J, 3330 Hospital Drive N.W. | Calgary, AB | T2N 4N1
Phone: (403) 210-6273

mri_robust_template --mov 2001A_T1.nii.gz 2001B_T1.nii.gz --lta 2001A2template.lta 2001B2template.lta --average 1 --template 2001_template.nii.gz --satit --debug

$Id: mri_robust_template.cpp,v 1.54 2016/05/05 21:17:08 mreuter Exp $

--mov: Using 2001A_T1.nii.gz as movable/source volume.

--mov: Using 2001B_T1.nii.gz as movable/source volume.

Total: 2 input volumes

--lta: Will output LTA transforms

--average: Using method 1 for template computation.

--template: Using 2001_template.nii.gz as template output volume.

--satit: Will estimate SAT iteratively!

--debug: Will output debug info and files!

reading source '2001A_T1.nii.gz'...

converting source '2001A_T1.nii.gz' to bspline ...

MRItoBSpline degree 3

reading source '2001B_T1.nii.gz'...

converting source '2001B_T1.nii.gz' to bspline ...

MRItoBSpline degree 3

Will use TP 1 as random initial target (seed 113817 ).

MultiRegistration::initializing Xforms (init 1 , maxres 0 , iterate 5 , epsit 0.01 ) :

MultiRegistration Parameters

Outdir: ./

TransOnly: false

Rigid: true

Robust: true

Satit: true

Iscale: false

IscaleOnly: false

NoMulti: false

SubsampleSize: -1

HighIt: -1

FixVoxel: false

KeepType: true

Average: 1

DoublePrec: false

BackupWeights: false

SampleType: 5

CRASCenter: false

Debug: 1

[init] ========================= TP 2 to TP 1 ==============================

to TP 1 ( 2001A_T1.nii.gz )

Registration::setSourceAndTarget(MRI s, MRI t, keeptype = TRUE )

Type Source : 4 Type Target : 4 keeping type

Mov: (1, 1, 1)mm and dim (256, 256, 176)

Dst: (1, 1, 1)mm and dim (256, 256, 176)

Asserting both images: 1mm isotropic

- no Mov reslice necessary

- no Dst reslice necessary

Registration::findSaturation

- computing centroids

- computing initial transform

-- using translation info

- Get Gaussian Pyramid Limits ( min size: 16 max size: -1 )

- Build Gaussian Pyramid ( Limits min steps: 0 max steps: 3 )

- Max Resolution used: 2

-- gpS ( 64 , 64 , 44 )

-- gpT ( 64 , 64 , 44 )

- running loop to estimate saturation parameter:

-- Iteration: 1 trying sat: 16

min sat: 0 ( -1 ), max sat: 16 ( 0.0798945 ), sat diff: 16, (wlimit=0.16)

-- Iteration: 2 trying sat: 8

min sat: 8 ( 0.17712 ), max sat: 16 ( 0.0798945 ), sat diff: 8, (wlimit=0.16)

-- Iteration: 3 trying sat: 12

min sat: 8 ( 0.17712 ), max sat: 12 ( 0.109687 ), sat diff: 4, (wlimit=0.16)

-- Iteration: 4 trying sat: 10

min sat: 8 ( 0.17712 ), max sat: 10 ( 0.138304 ), sat diff: 2, (wlimit=0.16)

-- Iteration: 5 trying sat: 9

min sat: 8 ( 0.17712 ), max sat: 9 ( 0.154011 ), sat diff: 1, (wlimit=0.16)

-- Iteration: 6 trying sat: 8.5

- final SAT: 9 ( it: 6 , weight check 0.154011 <= 0.16 )

Registration::computeMultiresRegistration

- computing centroids

- computing initial transform

-- using translation info

- Get Gaussian Pyramid Limits ( min size: 16 max size: -1 )

- initial transform:

Ti = [ ...

1.0000000000000 0 0 -1.2743032370447

0 1.0000000000000 0 6.3477877176776

0 0 1.0000000000000 1.3765177143265

0 0 0 1.0000000000000 ]

- initial iscale: Ii =1

Resolution: 3 S( 32 32 22 ) T( 32 32 22 )

Iteration(f): 1

-- diff. to prev. transform: 4.3252

Iteration(f): 2

-- diff. to prev. transform: 0.226503

Iteration(f): 3

-- diff. to prev. transform: 0.0809929

Iteration(f): 4

-- diff. to prev. transform: 0.0448288

Iteration(f): 5

-- diff. to prev. transform: 0.0306485 max it: 5 reached!

Resolution: 2 S( 64 64 44 ) T( 64 64 44 )

Iteration(f): 1

-- diff. to prev. transform: 1.21431

Iteration(f): 2

-- diff. to prev. transform: 0.153379

Iteration(f): 3

-- diff. to prev. transform: 0.108256

Iteration(f): 4

-- diff. to prev. transform: 0.0849185

Iteration(f): 5

-- diff. to prev. transform: 0.0709608 max it: 5 reached!

Resolution: 1 S( 128 128 88 ) T( 128 128 88 )

Iteration(f): 1

-- diff. to prev. transform: 0.803945

Iteration(f): 2

-- diff. to prev. transform: 0.19292

Iteration(f): 3

-- diff. to prev. transform: 0.124963

Iteration(f): 4

-- diff. to prev. transform: 0.0841509

Iteration(f): 5

-- diff. to prev. transform: 0.0593591 max it: 5 reached!

Resolution: 0 S( 256 256 176 ) T( 256 256 176 )

Iteration(f): 1

-- diff. to prev. transform: 1.56407

Iteration(f): 2

-- diff. to prev. transform: 0.1626

Iteration(f): 3

-- diff. to prev. transform: 0.0426001

Iteration(f): 4

-- diff. to prev. transform: 0.0296054

Iteration(f): 5

-- diff. to prev. transform: 0.0229654 max it: 5 reached!

- final transform:

Tf = [ ...

0.9991201336644 -0.0403162432589 0.0115567744604 13.1803922948678

0.0399069838938 0.9986352667380 0.0336903051890 -5.3133887637339

-0.0128992690853 -0.0331994662113 0.9993655008556 6.0924108135986

0 0 0 1.0000000000000 ]

- final iscale: If = 1

**********************************************************

* WARNING: Registration did not converge in 5 steps!

* Problem might be ill posed.

* Please inspect output manually!

**********************************************************

CentroidT = [ 144.3744389157496 140.7612889458485 89.5924668508406 1.0000000000000 ]

CentroidSinT = [ 154.3014274378032 137.7010981446322 87.9111681965244 1.0000000000000 ]

Centroid = [ 149.3379331767764 139.2311935452403 88.7518175236825 1.0000000000000 ]

* WARNING: TP 2 to 1 did not converge !!

meant = [ 1.1453379004720 -1.3465092209200 -3.1182479079224 ]

Input rotation's max deviation from rotation is: 0.000393412

MyMatrix::RotationMean input(s) not rotation