System and method for automatic registration of 4D (3D plus time) renal perfusion MRI data
View Patent ↗A method for registering digital renal perfusion images includes selecting a volume of interest (VOI) containing a kidney in a reference renal perfusion image, computing 3D intensity gradients for a plurality of points in the VOI of the reference renal perfusion image, computing 3D intensity gradients for a plurality of points in a search window of a current renal perfusion image, and maximizing a similarity measure between the reference image VOI and the current image search window, where the similarity measure is a function of the 3D intensity gradients computed for the reference image and the current image.
1. A computer implemented method for registering digital renal perfusion images, the method performed by the computer comprising the steps of:
selecting a volume of interest (VOI) containing a kidney in a reference renal perfusion image, wherein said volume of interest in said reference image is selected by two 2D bounding boxes in perpendicular planes;
computing 3D intensity gradients for a plurality of points in said VOI of said reference renal perfusion image;
computing 3D intensity gradients for a plurality of points in a search window of a current renal perfusion image; and
maximizing a similarity measure between the reference image VOI and the current image search window, wherein said similarity measure is a function of said 3D intensity gradients computed for said reference image and said current image.
2. The method of claim 1 , wherein registration is determined by translational displacement of the current image with respect to the reference image.
3. The method of claim 2 , wherein parameters associated with a maximal similarity measure determine the translational displacement of each point in the VOI of the reference image to a corresponding point in the current image.
4. The method of claim 3 , wherein maximizing the similarity measure between the reference image VOI and the current image search window comprises calculating said similarity measure for a plurality of translational displacements (dx,dy,dz) from points in said reference VOI to points in said current image search window, and selecting a displacement (dx*,dy*,dz*) that maximizes the similarity measure.
5. The method of claim 4 , wherein said similarity measure is defined as
S
(
dx
,
dy
,
dz
)
=
∑
(
x
,
y
,
z
)
∈
V
w
(
x
,
y
,
z
;
dx
,
dy
,
dz
)
v
r
(
x
,
y
,
z
)
T
v
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
,
wherein w(x,y,z;dx,dy,dz) is defined by
w
(
x
,
y
,
z
;
dx
,
dy
,
dz
)
=
M
r
(
x
,
y
,
z
)
M
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
∑
(
x
,
y
,
z
)
∈
V
M
r
(
x
,
y
,
z
)
M
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
,
wherein
M
r
(
x
,
y
,
z
)
=
(
∂
I
r
(
x
,
y
,
z
)
∂
x
)
2
+
(
∂
I
r
(
x
,
y
,
z
)
∂
y
)
2
+
(
∂
I
r
(
x
,
y
,
z
)
∂
z
)
2
is a magnitude of the 3D intensity gradients
∂
I
r
(
x
,
y
,
z
)
∂
x
,
∂
I
r
(
x
,
y
,
z
)
∂
y
,
∂
I
r
(
x
,
y
,
z
)
∂
z
of the reference image VOI I r , and
v
r
(
x
,
y
,
z
)
=
1
M
r
(
x
,
y
,
z
)
[
∂
I
r
(
x
,
y
,
z
)
∂
x
,
∂
I
r
(
x
,
y
,
z
)
∂
y
,
∂
I
r
(
x
,
y
,
z
)
∂
z
]
T
is a unit orientation vector of the gradient, and wherein the sum is over all points in the VOI.
6. The method of claim 5 , further comprising restricting said sum to a subset of the VOI comprising those points whose intensity gradient magnitudes are above a threshold.
7. The method of claim 6 , wherein the threshold is set to select about 15% of the points in the VOI.
8. The method of claim 1 , wherein said current image search window is determined based on a location of the kidney in the reference image VOI and its maximum possible displacements.
9. A computer implemented method for registering digital renal perfusion images, the method performed by the computer comprising the steps of:
receiving a sequence of digital renal perfusion images;
selecting a volume of interest (VOI) containing a kidney in a reference renal perfusion image in said sequence of images;
building a template from 3D intensity gradients of a subset of points in said reference image VOI, wherein building the template comprises selecting those points in said VOI whose intensity gradient magnitudes are above a threshold, wherein each selected point is associated with its corresponding intensity gradient magnitude and intensity gradient orientation vector; and
calculating translational displacements at corresponding points in a current image in said sequence of images that maximize a similarity measure between the current image and the template.
10. The method of claim 9 , further comprising computing 3D intensity gradients for a plurality of points in a search window of the current renal perfusion image.
11. The method of claim 10 , wherein said similarity measure favors those current image gradients with large magnitudes and whose orientations are consistent with those of stronger edges in the reference VOI, and is invariant to rapidly changing image contrast.
12. The method of claim 11 , wherein said similarity measure is defined as
S
~
(
dx
,
dy
,
dz
)
=
∑
(
x
,
y
,
z
)
∈
V
1
w
~
(
x
,
y
,
z
;
dx
,
dy
,
dz
)
v
r
(
x
,
y
,
z
)
T
v
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
.
wherein {tilde over (w)}(x,y,z;dx,dy,dz) is defined by
w
~
(
x
,
y
,
z
;
dx
,
dy
,
dz
)
=
M
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
∑
(
x
,
y
,
z
)
∈
V
1
M
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
,
wherein
M
r
(
x
,
y
,
z
)
=
(
∂
I
r
(
x
,
y
,
z
)
∂
x
)
2
+
(
∂
I
r
(
x
,
y
,
z
)
∂
y
)
2
+
(
∂
I
r
(
x
,
y
,
z
)
∂
z
)
2
is a magnitude of the 3D intensity gradients
∂
I
r
(
x
,
y
,
z
)
∂
x
,
∂
I
r
(
x
,
y
,
z
)
∂
y
,
∂
I
r
(
x
,
y
,
z
)
∂
z
of the reference image VOI I r , and
v
r
(
x
,
y
,
z
)
=
1
M
r
(
x
,
y
,
z
)
[
∂
I
r
(
x
,
y
,
z
)
∂
x
,
∂
I
r
(
x
,
y
,
z
)
∂
y
,
∂
I
r
(
x
,
y
,
z
)
∂
z
]
T
is a unit orientation vector of the gradient, V 1 is the template, and wherein the sum is over all points in the V 1 .
13. The method of claim 10 , wherein maximizing said similarity measure between the reference image VOI and the current image search window comprises calculating said similarity measure for a plurality of translational displacements (dx,dy,dz) from points in said reference VOI template to points in said current image search window, and selecting a displacement (dx*,dy*,dz*) that maximizes the similarity measure.
14. A non-transitory program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform the method steps for registering digital renal perfusion images, the method comprising the steps of:
selecting a volume of interest (VOI) containing a kidney in a reference renal perfusion image;
computing 3D intensity gradients for a plurality of points in said VOI of said reference renal perfusion image;
computing 3D intensity gradients for a plurality of points in a search window of a current renal perfusion image, wherein said current image search window is determined based on a location of the kidney in the reference image VOI and its maximum possible displacements; and
maximizing a similarity measure between the reference image VOI and the current image search window, wherein said similarity measure is a function of said 3D intensity gradients computed for said reference image and said current image.
15. The computer readable program storage device of claim 14 , wherein said volume of interest in said reference image is selected by two 2D bounding boxes in perpendicular planes.
16. The computer readable program storage device of claim 14 , wherein registration is determined by translational displacement of the current image with respect to the reference image.
17. The computer readable program storage device of claim 16 , wherein parameters associated with a maximal similarity measure determine the translational displacement of each point in the VOI of the reference image to a corresponding point in the current image.
18. The computer readable program storage device of claim 17 , wherein maximizing the similarity measure between the reference image VOI and the current image search window comprises calculating said similarity measure for a plurality of translational displacements (dx,dy,dz) from points in said reference VOI to points in said current image search window, and selecting a displacement dx*,dy*,dz*) that maximizes the similarity measure.
19. The computer readable program storage device of claim 18 , wherein said similarity measure is defined as
S
(
dx
,
dy
,
dz
)
=
∑
(
x
,
y
,
z
)
∈
V
w
(
x
,
y
,
z
;
dx
,
dy
,
dz
)
v
r
(
x
,
y
,
z
)
T
v
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
,
wherein w(x,y,z;dx,dy,dz) is defined by
w
(
x
,
y
,
z
;
dx
,
dy
,
dz
)
=
M
r
(
x
,
y
,
z
)
M
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
∑
(
x
,
y
,
z
)
∈
V
M
r
(
x
,
y
,
z
)
M
c
(
x
+
dx
,
y
+
dy
,
z
+
dz
)
,
wherein
M
r
(
x
,
y
,
z
)
=
(
∂
I
r
(
x
,
y
,
z
)
∂
x
)
2
+
(
∂
I
r
(
x
,
y
,
z
)
∂
y
)
2
+
(
∂
I
r
(
x
,
y
,
z
)
∂
z
)
2
is a magnitude of the 3D intensity gradients
∂
I
r
(
x
,
y
,
z
)
∂
x
,
∂
I
r
(
x
,
y
,
z
)
∂
y
,
∂
I
r
(
x
,
y
,
z
)
∂
z
of the reference image VOI I r , and
v
r
(
x
,
y
,
z
)
=
1
M
r
(
x
,
y
,
z
)
[
∂
I
r
(
x
,
y
,
z
)
∂
x
,
∂
I
r
(
x
,
y
,
z
)
∂
y
,
∂
I
r
(
x
,
y
,
z
)
∂
z
]
T
is a unit orientation vector of the gradient, and wherein the sum is over all points in the VOI.
20. The computer readable program storage device of claim 19 , the method further comprising restricting said sum to a subset of the VOI comprising those points whose intensity gradient magnitudes are above a threshold.
21. The computer readable program storage device of claim 20 , wherein the threshold is set to select about 15% of the points in the VOI.