Skip to content
GitLab
Explore
Sign in
Primary navigation
Search or go to…
Project
F
fdt
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Snippets
Build
Pipelines
Jobs
Pipeline schedules
Artifacts
Deploy
Releases
Container Registry
Model registry
Operate
Environments
Monitor
Incidents
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
FSL
fdt
Commits
c1752957
Commit
c1752957
authored
17 years ago
by
Saad Jbabdi
Browse files
Options
Downloads
Patches
Plain Diff
Vector registration v1.0
parent
47964802
No related branches found
Branches containing commit
No related tags found
Tags containing commit
No related merge requests found
Changes
2
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
vecreg.cc
+334
-0
334 additions, 0 deletions
vecreg.cc
vecreg.h
+23
-0
23 additions, 0 deletions
vecreg.h
with
357 additions
and
0 deletions
vecreg.cc
0 → 100755
+
334
−
0
View file @
c1752957
/* vector_flirt.cc
Saad Jbabdi, FMRIB Image Analysis Group
Copyright (C) 2007 University of Oxford */
/* CCOPYRIGHT */
#include
"vecreg.h"
#include
"utils/options.h"
using
namespace
Utilities
;
string
title
=
"vector_flirt (Version 1.0)
\n
Vector Affine/non linear Tranformation with Orientation Preservation"
;
string
examples
=
"vector_flirt -i <input4Dvector> -o <output4D> -t <transformation>"
;
Option
<
bool
>
verbose
(
string
(
"-v,--verbose"
),
false
,
string
(
"switch on diagnostic messages"
),
false
,
no_argument
);
Option
<
bool
>
help
(
string
(
"-h,--help"
),
false
,
string
(
"display this message"
),
false
,
no_argument
);
Option
<
string
>
ivector
(
string
(
"-i,--input"
),
string
(
""
),
string
(
"filename of input vector"
),
true
,
requires_argument
);
Option
<
string
>
ovector
(
string
(
"-o,--output"
),
string
(
""
),
string
(
"filename of output registered vector"
),
true
,
requires_argument
);
Option
<
string
>
ref
(
string
(
"-r,--ref"
),
string
(
""
),
string
(
"filename of reference (target) volume"
),
true
,
requires_argument
);
Option
<
string
>
matrix
(
string
(
"-t,--affine"
),
string
(
""
),
string
(
"filename of affine transformation matrix"
),
false
,
requires_argument
);
Option
<
string
>
warp
(
string
(
"-w,--warpfield"
),
string
(
""
),
string
(
"filename of 4D warp field for nonlinear registration"
),
false
,
requires_argument
);
Option
<
string
>
interpmethod
(
string
(
"--interp"
),
string
(
"nearestneighbour"
),
string
(
"interpolation method : nearestneighbour (default) or trilinear or sinc"
),
false
,
requires_argument
);
Option
<
string
>
maskfile
(
string
(
"-m,--mask"
),
string
(
""
),
string
(
"brain mask in input space"
),
false
,
requires_argument
);
////////////////////////////////////////////////////////
void
vecreg_aff
(
const
volume4D
<
float
>&
tens
,
volume4D
<
float
>&
oV1
,
const
volume
<
float
>&
refvol
,
const
Matrix
&
M
,
const
volume
<
float
>&
mask
){
Matrix
iM
(
4
,
4
),
R
(
3
,
3
);
iM
=
M
.
i
();
// extract rotation matrix from M
Matrix
F
(
3
,
3
),
u
(
3
,
3
),
v
(
3
,
3
);
DiagonalMatrix
d
(
3
);
F
=
M
.
SubMatrix
(
1
,
3
,
1
,
3
);
SVD
(
F
*
F
.
t
(),
d
,
u
,
v
);
R
=
(
u
*
sqrt
(
d
)
*
v
.
t
()).
i
()
*
F
;
ColumnVector
seeddim
(
3
),
targetdim
(
3
);
seeddim
<<
tens
.
xdim
()
<<
tens
.
ydim
()
<<
tens
.
zdim
();
targetdim
<<
refvol
.
xdim
()
<<
refvol
.
ydim
()
<<
refvol
.
zdim
();
SymmetricMatrix
Tens
(
3
);
ColumnVector
X_seed
(
3
),
X_target
(
3
);
ColumnVector
V_seed
(
3
),
V_target
(
3
);
for
(
int
z
=
0
;
z
<
oV1
.
zsize
();
z
++
)
for
(
int
y
=
0
;
y
<
oV1
.
ysize
();
y
++
)
for
(
int
x
=
0
;
x
<
oV1
.
xsize
();
x
++
){
// compute seed coordinates
X_target
<<
x
<<
y
<<
z
;
X_seed
=
vox_to_vox
(
X_target
,
targetdim
,
seeddim
,
iM
);
if
(
mask
((
int
)
X_seed
(
1
),(
int
)
X_seed
(
2
),(
int
)
X_seed
(
3
))
==
0
){
continue
;
}
// compute interpolated tensor
Tens
<<
tens
[
0
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
1
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
2
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
3
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
4
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
5
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
));
// compute first eigenvector
EigenValues
(
Tens
,
d
,
v
);
V_seed
=
v
.
Column
(
3
);
// rotate vector
V_target
=
R
*
V_seed
;
V_target
/=
sqrt
(
V_target
.
SumSquare
());
oV1
(
x
,
y
,
z
,
0
)
=
V_target
(
1
);
oV1
(
x
,
y
,
z
,
1
)
=
V_target
(
2
);
oV1
(
x
,
y
,
z
,
2
)
=
V_target
(
3
);
}
}
void
sjgradient
(
const
volume
<
float
>&
im
,
volume4D
<
float
>&
grad
){
grad
.
reinitialize
(
im
.
xsize
(),
im
.
ysize
(),
im
.
zsize
(),
3
);
copybasicproperties
(
im
,
grad
[
0
]);
int
fx
,
fy
,
fz
,
bx
,
by
,
bz
;
float
dx
,
dy
,
dz
;
for
(
int
z
=
0
;
z
<
grad
.
zsize
();
z
++
){
fz
=
z
==
(
grad
.
zsize
()
-
1
)
?
0
:
1
;
bz
=
z
==
0
?
0
:
-
1
;
dz
=
(
fz
==
0
||
bz
==
0
)
?
1.0
:
2.0
;
for
(
int
y
=
0
;
y
<
grad
.
ysize
();
y
++
){
fy
=
y
==
(
grad
.
ysize
()
-
1
)
?
0
:
1
;
by
=
y
==
0
?
0
:
-
1
;
dy
=
(
fy
==
0
||
by
==
0
)
?
1.0
:
2.0
;
for
(
int
x
=
0
;
x
<
grad
.
xsize
();
x
++
){
fx
=
x
==
(
grad
.
xsize
()
-
1
)
?
0
:
1
;
bx
=
x
==
0
?
0
:
-
1
;
dx
=
(
fx
==
0
||
bx
==
0
)
?
1.0
:
2.0
;
grad
[
0
](
x
,
y
,
z
)
=
(
im
(
x
+
fx
,
y
,
z
)
-
im
(
x
+
bx
,
y
,
z
))
/
dx
;
grad
[
1
](
x
,
y
,
z
)
=
(
im
(
x
,
y
+
fy
,
z
)
-
im
(
x
,
y
+
by
,
z
))
/
dy
;
grad
[
2
](
x
,
y
,
z
)
=
(
im
(
x
,
y
,
z
+
fz
)
-
im
(
x
,
y
,
z
+
bz
))
/
dz
;
}
}
}
}
void
vecreg_nonlin
(
const
volume4D
<
float
>&
tens
,
volume4D
<
float
>&
oV1
,
const
volume
<
float
>&
refvol
,
volume4D
<
float
>&
warp
,
const
volume
<
float
>&
mask
){
ColumnVector
X_seed
(
3
),
X_target
(
3
);
//float dxx=tens.xdim(),dyy=tens.ydim(),dzz=tens.zdim();
float
dx
=
oV1
.
xdim
(),
dy
=
oV1
.
ydim
(),
dz
=
oV1
.
zdim
();
//float nxx=(float)tens.xsize()/2.0,nyy=(float)tens.ysize()/2.0,nzz=(float)tens.zsize()/2.0;
//float nx=(float)oV1.xsize()/2.0,ny=(float)oV1.ysize()/2.0,nz=(float)oV1.zsize()/2.0;
// transform mm warp to voxel warp
// the warpfield here has been transfomed by MJ's script
for
(
int
z
=
0
;
z
<
warp
[
0
].
zsize
();
z
++
)
for
(
int
y
=
0
;
y
<
warp
[
0
].
ysize
();
y
++
)
for
(
int
x
=
0
;
x
<
warp
[
0
].
xsize
();
x
++
){
warp
[
0
](
x
,
y
,
z
)
/=
dx
;
warp
[
1
](
x
,
y
,
z
)
/=
dy
;
warp
[
2
](
x
,
y
,
z
)
/=
dz
;
}
// compute transformation jacobian
volume4D
<
float
>
jx
(
mask
.
xsize
(),
mask
.
ysize
(),
mask
.
zsize
(),
3
);
volume4D
<
float
>
jy
(
mask
.
xsize
(),
mask
.
ysize
(),
mask
.
zsize
(),
3
);
volume4D
<
float
>
jz
(
mask
.
xsize
(),
mask
.
ysize
(),
mask
.
zsize
(),
3
);
sjgradient
(
warp
[
0
],
jx
);
sjgradient
(
warp
[
1
],
jy
);
sjgradient
(
warp
[
2
],
jz
);
ColumnVector
V_seed
(
3
),
V_target
(
3
);
Matrix
R
(
3
,
3
),
I
(
3
,
3
);
I
<<
1
<<
0
<<
0
<<
0
<<
1
<<
0
<<
0
<<
0
<<
1
;
Matrix
F
(
3
,
3
),
Jw
(
3
,
3
),
u
(
3
,
3
),
v
(
3
,
3
);
DiagonalMatrix
d
(
3
);
SymmetricMatrix
Tens
(
3
);
for
(
int
z
=
0
;
z
<
oV1
.
zsize
();
z
++
)
for
(
int
y
=
0
;
y
<
oV1
.
ysize
();
y
++
)
for
(
int
x
=
0
;
x
<
oV1
.
xsize
();
x
++
){
X_seed
<<
round
(
x
+
warp
[
0
](
x
,
y
,
z
))
<<
round
(
y
+
warp
[
1
](
x
,
y
,
z
))
<<
round
(
z
+
warp
[
2
](
x
,
y
,
z
));
if
(
mask
((
int
)
X_seed
(
1
),(
int
)
X_seed
(
2
),(
int
)
X_seed
(
3
))
==
0
){
continue
;
}
// compute interpolated tensor
Tens
<<
tens
[
0
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
1
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
2
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
3
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
4
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
))
<<
tens
[
5
].
interpolate
(
X_seed
(
1
),
X_seed
(
2
),
X_seed
(
3
));
// compute first eigenvector
EigenValues
(
Tens
,
d
,
v
);
V_seed
=
v
.
Column
(
3
);
// Local Jacobian of the backward warpfield
Jw
<<
jx
(
x
,
y
,
z
,
0
)
<<
jx
(
x
,
y
,
z
,
1
)
<<
jx
(
x
,
y
,
z
,
2
)
<<
jy
(
x
,
y
,
z
,
0
)
<<
jy
(
x
,
y
,
z
,
1
)
<<
jy
(
x
,
y
,
z
,
2
)
<<
jz
(
x
,
y
,
z
,
0
)
<<
jz
(
x
,
y
,
z
,
1
)
<<
jz
(
x
,
y
,
z
,
2
);
// compute local forward affine transformation
F
=
(
I
+
Jw
).
i
();
// reorient according to affine reorientation scheme
SVD
(
F
*
F
.
t
(),
d
,
u
,
v
);
R
=
(
u
*
sqrt
(
d
)
*
v
.
t
()).
i
()
*
F
;
//R=ppd(F,V_seed);
V_target
=
R
*
V_seed
;
V_target
/=
sqrt
(
V_target
.
SumSquare
());
oV1
(
x
,
y
,
z
,
0
)
=
V_target
(
1
);
oV1
(
x
,
y
,
z
,
1
)
=
V_target
(
2
);
oV1
(
x
,
y
,
z
,
2
)
=
V_target
(
3
);
}
}
int
do_vecreg
(){
volume4D
<
float
>
ivol
,
warpvol
;
volume
<
float
>
refvol
,
mask
;
Matrix
Aff
(
4
,
4
);
volumeinfo
vinfo
;
if
((
matrix
.
set
())){
read_ascii_matrix
(
Aff
,
matrix
.
value
());
}
if
((
warp
.
set
())){
if
(
verbose
.
value
())
cerr
<<
"Loading warpfield"
<<
endl
;
read_volume4D
(
warpvol
,
warp
.
value
());
}
if
(
verbose
.
value
())
cerr
<<
"Loading volumes"
<<
endl
;
read_volume4D
(
ivol
,
ivector
.
value
());
read_volume
(
refvol
,
ref
.
value
(),
vinfo
);
volume4D
<
float
>
ovol
(
refvol
.
xsize
(),
refvol
.
ysize
(),
refvol
.
zsize
(),
3
);
copybasicproperties
(
refvol
,
ovol
);
// set interpolation method
if
(
interpmethod
.
value
()
==
"nearestneighbour"
)
ivol
.
setinterpolationmethod
(
nearestneighbour
);
else
if
(
interpmethod
.
value
()
==
"sinc"
)
ivol
.
setinterpolationmethod
(
sinc
);
else
ivol
.
setinterpolationmethod
(
trilinear
);
if
(
maskfile
.
value
()
!=
""
)
read_volume
(
mask
,
maskfile
.
value
());
else
{
copybasicproperties
(
ivol
,
mask
);
mask
=
1
;
}
///////////////////////
// tensor for interpolation
volume4D
<
float
>
tens
(
ivol
.
xsize
(),
ivol
.
ysize
(),
ivol
.
zsize
(),
6
);
copybasicproperties
(
ivol
,
tens
);
for
(
int
z
=
0
;
z
<
ivol
.
zsize
();
z
++
)
for
(
int
y
=
0
;
y
<
ivol
.
ysize
();
y
++
)
for
(
int
x
=
0
;
x
<
ivol
.
xsize
();
x
++
){
tens
(
x
,
y
,
z
,
0
)
=
ivol
(
x
,
y
,
z
,
0
)
*
ivol
(
x
,
y
,
z
,
0
);
tens
(
x
,
y
,
z
,
1
)
=
ivol
(
x
,
y
,
z
,
1
)
*
ivol
(
x
,
y
,
z
,
0
);
tens
(
x
,
y
,
z
,
2
)
=
ivol
(
x
,
y
,
z
,
1
)
*
ivol
(
x
,
y
,
z
,
1
);
tens
(
x
,
y
,
z
,
3
)
=
ivol
(
x
,
y
,
z
,
2
)
*
ivol
(
x
,
y
,
z
,
0
);
tens
(
x
,
y
,
z
,
4
)
=
ivol
(
x
,
y
,
z
,
2
)
*
ivol
(
x
,
y
,
z
,
1
);
tens
(
x
,
y
,
z
,
5
)
=
ivol
(
x
,
y
,
z
,
2
)
*
ivol
(
x
,
y
,
z
,
2
);
}
//time_t _time=time(NULL);
if
(
matrix
.
set
()){
if
(
verbose
.
value
())
cerr
<<
"Affine registration"
<<
endl
;
vecreg_aff
(
tens
,
ovol
,
refvol
,
Aff
,
mask
);
}
else
{
if
(
verbose
.
value
())
cerr
<<
"Nonlinear registration"
<<
endl
;
vecreg_nonlin
(
tens
,
ovol
,
refvol
,
warpvol
,
mask
);
}
//cout<<"elapsed time:"<<time(NULL)-_time<<" sec"<<endl;
save_volume4D
(
ovol
,
ovector
.
value
());
return
0
;
}
int
main
(
int
argc
,
char
*
argv
[]){
Tracer
tr
(
"main"
);
OptionParser
options
(
title
,
examples
);
try
{
options
.
add
(
verbose
);
options
.
add
(
help
);
options
.
add
(
ivector
);
options
.
add
(
ovector
);
options
.
add
(
ref
);
options
.
add
(
matrix
);
options
.
add
(
warp
);
options
.
add
(
interpmethod
);
options
.
add
(
maskfile
);
options
.
parse_command_line
(
argc
,
argv
);
if
(
(
help
.
value
())
||
(
!
options
.
check_compulsory_arguments
(
true
))
){
options
.
usage
();
exit
(
EXIT_FAILURE
);
}
if
(
(
matrix
.
set
())
&&
(
warp
.
set
())
){
options
.
usage
();
cerr
<<
endl
<<
"Cannot specify both --affine AND --warpfield"
<<
endl
<<
endl
;
exit
(
EXIT_FAILURE
);
}
if
(
(
matrix
.
unset
())
&&
(
warp
.
unset
())
){
options
.
usage
();
cerr
<<
endl
<<
"Please Specify either --affine OR --warpfield"
<<
endl
<<
endl
;
exit
(
EXIT_FAILURE
);
}
}
catch
(
X_OptionError
&
e
)
{
options
.
usage
();
cerr
<<
endl
<<
e
.
what
()
<<
endl
;
exit
(
EXIT_FAILURE
);
}
catch
(
std
::
exception
&
e
)
{
cerr
<<
e
.
what
()
<<
endl
;
}
return
do_vecreg
();
}
This diff is collapsed.
Click to expand it.
vecreg.h
0 → 100755
+
23
−
0
View file @
c1752957
#include
<cmath>
#include
<stdlib.h>
#include
"newimage/newimageall.h"
#include
"miscmaths/miscmaths.h"
using
namespace
NEWIMAGE
;
using
namespace
NEWMAT
;
using
namespace
std
;
ReturnMatrix
rodrigues
(
const
float
&
,
ColumnVector
&
);
ReturnMatrix
rodrigues
(
const
float
&
,
const
float
&
,
ColumnVector
&
);
ReturnMatrix
rodrigues
(
const
ColumnVector
&
,
const
ColumnVector
&
);
ReturnMatrix
ppd
(
const
Matrix
&
,
const
ColumnVector
&
,
const
ColumnVector
&
);
void
vecreg_aff
(
const
volume4D
<
float
>&
,
volume4D
<
float
>&
,
const
volume
<
float
>&
,
const
Matrix
&
,
const
volume
<
float
>&
);
void
vecreg_nonlin
(
const
volume4D
<
float
>&
,
volume4D
<
float
>&
,
const
volume
<
float
>&
,
volume4D
<
float
>&
,
const
volume
<
float
>&
);
void
sjgradient
(
const
volume
<
float
>&
,
volume4D
<
float
>&
);
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment