Non-rigid registration between CT images and ultrasound images
There are disclosed embodiments for non-rigid image registration between 3-dimensional ultrasound and CT images by using intensity and gradient information of vessels and diaphragm. An ultrasound image forming unit transmits/receives ultrasound signals to/from a target object to thereby output electrical receive signals, and forms 3-dimensional ultrasound images based on the electrical receive signals. A CT image forming unit forms 3-dimensional CT images of the target object. A registration unit determines first and second objective functions associated with diaphragm and vessel regions of the target object, respectively, based on intensity and gradient information upon portions corresponding to the diaphragm and vessel regions in each of the 3-dimensional ultrasound and CT images. The registration unit performs non-rigid image registration between the 3-dimensional ultrasound images and the 3-dimensional CT images based on the first and second objective functions.
1. A system for a non-rigid image registration between ultrasound images and computerized tomography (CT) images, comprising:
an ultrasound image forming unit configured to transmit/receive ultrasound signals to/from a target object having a first region and a second region to thereby output electrical receive signals, the ultrasound image forming unit being further configured to form 3-dimensional ultrasound images based on the electrical receive signals;
a CT image forming unit configured to form 3-dimensional CT images of the target object; and
a registration unit configured to measure a first statistical entropy based on intensity values in the first region of the 3-dimensional ultrasound images, gradient magnitudes in the first region of the CT images and an edge orientation coincidence function in the first region to determine a first objective function and to measure a second statistical entropy based on intensity values in the second region in each of the 3-dimensional ultrasound and CT images and an edge orientation coincidence function in the second region to determine a second objective function, the registration unit being further configured to perform the non-rigid image registration between the 3-dimensional ultrasound images and the 3-dimensional CT images based on the first and second objective functions.
2. The system of claim 1 , wherein the 3-dimensional ultrasound and CT images are 3-dimensional ultrasound and CT liver images, and wherein the first and second regions are diaphragm and vessel regions, respectively.
3. The system of claim 2 , wherein the registration unit is configured to:
perform an affine registration between the 3-dimensional ultrasound images and the 3-dimensional CT images;
model a local deformation appearing at the affine-registered 3-dimensional ultrasound and CT images to define a transformation parameter;
define a plurality of regions of interest (ROIs) in the 3-dimensional CT images;
determine the first objective function in the diaphragm region and the second objective function in the vessel region within the ROIs, define a cost function based on the first and second objective functions;
perform optimization for updating the transformation parameter based on the cost function; and
form transformed CT images based on the optimized transformation parameter and perform non-rigid image registration upon the transformed CT images and the ultrasound images.
4. The system of claim 3 , wherein the affine registration is carried out based on Iterative Closest Point.
5. The system of claim 3 , wherein the local deformation is modeled by imposing B-spline free foam deformation on the affine-registered 3-dimensional ultrasound and CT images.
6. The system of claim 3 , wherein the first and second objective functions are formed based on a 3-dimensional joint histogram based on edges orientation angles.
7. The system of claim 6 , wherein the ROIs are formed by segmenting vessels and tissues with a region-growing scheme and expanding edge regions of segments with a predetermined margin.
8. The system of claim 6 , wherein the cost function is defined by using first and second objective functions C diaphragm and C vessel as following equation:
C=C vessel +C diaphragm +λ·C smooth
wherein λ is a weighting value representing tradeoff between alignment of two images and smoothness of the transformation C smooth .
9. The system of claim 8 , wherein the transformation parameter Φ is updated by repeatedly performing Φ k+1 =Φ k +μ·∇C until |C k −C k−1 |<ε is satisfied, wherein μ denotes a block size, k is a natural number representing a repetition status of gradient descent and ε is a predetermined small positive value.