2D/3D image registration method
A method ( 100 ) that registers a 3D heart volume ( 112, 114 ) obtained from either a pre-operative MR image or CT image ( 102 ) to an intra-operative fluoroscopic image using a mesh of the heart structure ( 106 ) as the basis for the registration.
1. A method of registering a 2D image of an object to a 3D image of the object, comprising
forming a 3D mesh of the object from the 3D image,
translating the 3D image or the 3D mesh into a counterpart 2D image, and
spatially aligning the 3D mesh with the 2D image using the counterpart 2D image, including applying a cost function to evaluate and optimize a Chan-Vese-like functional used as an image dissimilarity metric.
2. The method of claim 1 , wherein the translating step comprises projecting the 3D mesh into two dimensions to obtain a 2D mask.
3. The method of claim 1 , wherein the spatially aligning step comprises optimizing a pose of the 3D mesh within the 2D image using the counterpart 2D image.
4. The method of claim 1 , wherein the translating and spatially aligning steps are iteratively performed so that an alignment distance between the 2D image and the 3D mesh reaches a predetermined amount.
5. The method of claim 1 , the spatially aligning step further comprises minimizing the image dissimilarity metric using a Downhill Simplex optimization method.
6. The method of claim 1 , further comprising obtaining a plurality of image frames for the 2D image and determining the optimal image frame for an initial spatial alignment of the 3D mesh with the 2D image.
7. A method of visualizing an object relative to an anatomical region of interest of a patient during a medical procedure, comprising:
a. obtaining an intra-operative 2D image of the anatomical region of interest;
b. obtaining volumetric imaging data of the anatomical region of interest;
c. obtaining a 3D mesh of the volumetric imaging data;
d. obtaining an estimated pose of the anatomical region of interest from the 3D mesh;
e. optimizing the estimated pose of the 3D mesh within the intra-operative 2D image to register the 3D mesh with the intra-operative 2D image and to localize the object relative to the anatomical region of interest;
f. obtaining a plurality of image frames for the intra-operative 2D image; and
g. detecting changes in the average intensity of the image in the plurality of image frames to determine an optimal image frame for an initial alignment of the 3D mesh with the intra-operative 2D image.
8. The method of claim 7 , wherein obtaining volumetric imaging data comprises acquiring 3D images of the anatomical region of interest using either CT or MR imaging.
9. The method of claim 7 , wherein obtaining a 3D mesh comprises segmenting the volume of the anatomical region of interest.
10. The method of claim 7 , wherein obtaining an estimated pose comprises computing and rendering a 2D mask from the 3D mesh.
11. The method of claim 10 , wherein the optimizing step comprises updating the estimated pose of the 3D mesh within the intra-operative 2D image using the 2D mask.
12. The method of claim 11 , wherein the optimizing step comprises calculating the distance between the intra-operative 2D image and the 2D mask.
13. The method of claim 11 , wherein the optimizing step comprises applying a cost function to evaluate and optimize a Chan-Vese-like functional used as an image dissimilarity metric.
14. The method of claim 13 , the optimizing step comprises minimizing the image dissimilarity metric using a Downhill Simplex optimization method.
15. The method of claim 11 , wherein the optimizing step comprises applying a cost function to evaluate and optimize an image dissimilarity metric that has the form of:
F
=
[
∫
x
∈
Ω
(
P
(
V
)
)
(
I
(
x
)
-
1
∫
y
∈
Ω
(
P
(
V
)
)
1
·
ⅆ
y
∫
y
∈
Ω
(
P
(
V
)
)
I
(
y
)
ⅆ
y
)
2
ⅆ
x
]
+
[
∫
x
∉
Ω
(
P
(
V
)
)
(
I
(
x
)
-
1
∫
y
∉
Ω
(
P
(
V
)
)
1
·
ⅆ
y
∫
y
∉
Ω
(
P
(
V
)
)
I
(
y
)
ⅆ
y
)
2
ⅆ
x
]
,
where I is a respective intra-operative 2D image, specifies the non-zero region in the intra-operative 2D image, P is the re-projection operator/DRR generator for the 2D mask, and V is the volume of the anatomical region of interest in the respective volumetric imaging data.
16. The method of claim 15 , the optimizing step comprises minimizing the image dissimilarity metric using a Downhill Simplex optimization method.
17. The method of claim 11 , wherein the obtaining an estimated pose and the optimizing step are iteratively performed so that an alignment distance between the intra-operative 2D image and the 2D mask reaches a predetermined amount.
18. The method of claim 7 , wherein obtaining an estimated pose comprises projecting the 3D mesh so as to render the 3D mesh in two dimensions.
19. The method of claim 18 , wherein projecting comprises applying a geometric transform on the 3D mesh.
20. The method of claim 18 , wherein the optimizing step comprises comparing the positional difference between the rendering of the projection of the 3D mesh and the intra-operative 2D image, which are overlaid one another.
21. The method of claim 20 , wherein comparing comprises applying a cost function to evaluate and optimize an image dissimilarity measure for finding a pose that registers the 3D mesh with the 2D image.
22. The method of claim 7 , further comprising
obtaining a plurality of image frames for the intra-operative 2D image and
determining an optimal image frame for an initial alignment of the 3D mesh with the intra-operative 2D image.
23. The method of claim 7 , wherein obtaining an intra-operative 2D image of the anatomical region of interest comprises obtaining an intra-operative 2D image of the anatomical region of interest that is enhanced by an imaging contrast agent received by the patient.
24. A system for providing multi-modal 2D/3D image registration during a medical procedure, comprising
an imager that acquires 2D image data of an anatomical region of a patient and
a processor that manipulates the acquired image data and stored 3D image data of the anatomical region to align a stored 3D image to a 2D image and to automatically track the anatomical region position during the medical procedure,
said processor manipulating the 3D image data, in part, to render the stored 3D image in two dimensions for a positional comparison between the 3D image and the 2D image,
wherein the processor applies a cost function to evaluate and optimize a Chan-Vese-like functional used as an image dissimilarity metric and an optimization method to minimize the image dissimilarity metric.
25. The system of claim 24 , wherein said processor has a graphics processing element that accelerates the manipulation of the 3D image data to render the stored 3D image in two dimensions.