Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
2000-09-15
2004-04-27
Au, Amelia M. (Department: 2621)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S128000, C382S131000
Reexamination Certificate
active
06728424
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to image processing systems and methods and, more particularly, to image registration systems that combine two or more images into a composite image. The present invention finds particular application in the field of medical imaging, however, it will be appreciated that the present invention is also applicable to other types of imaging systems wherein multiple images are correlated and combined into a composite image.
The term “image,” as used herein, refers to a three-dimensional (3D) or volumetric image, unless otherwise indicated. The acquisition of volume images via a variety of imaging modalities is well known in the medical field. Such modalities include, for example, magnetic resonance imaging is (MRI) techniques, x-ray computed tomography (CT), nuclear imaging techniques such as positron emission tomography (PET) and single photon emission computed tomography (SPECT), ultrasound, and so forth. Volume images so acquired are typically stored digitally, e.g., in a computer memory, as arrays of voxel values. Each voxel is associated with a location in 3D space (e.g., x, y, and z coordinates), and is assigned a color value, typically a gray scale intensity value.
Image fusion, or the combination of multiple associated images to form a composite image integrating the data therefrom, is often desirable in a clinical setting. In many cases, combined images might provide insights to the diagnostician that could not be obtained by viewing the images separately. Multi-modality image fusion is often useful since different imaging modalities provide information that tends to be complimentary in nature. For example, computed tomography (CT) and magnetic resonance (MR) imaging primarily provide anatomic or structural information while single photon emission computed tomography (SPECT) and positron emission tomography (PET) provide functional and metabolic information. The combination of a functional or metabolic image with a structural or anatomical image aids in localizing the functional image, thus improving diagnostic accuracy. For example, in the area of oncology, precise positioning of localization of functional images enables a clinician to assess lesion progression and/or treatment effectiveness. Also, such diagnostic studies are used in surgical and/or radiotherapeutic planning, where precise positioning is necessary to minimize the effect on healthy cells surrounding the target cells. It is also desirable at times to combine images from the same modality. For example, it may be desirable to combine the results of multiple MR scans, such as an MR angiograph, a contrast-enhanced MR image, or a functional MRI (fMRI) image, with another MR image, such as an anatomical MR image.
For the meaningful integration of data from multiple images, it is important that the images be properly registered. Image registration involves bringing the images into spatial alignment such that they are unambiguously linked together. A number of image registration techniques are known in the art.
One image registration technique requires that an individual with expertise in the structure of the object represented in the images label a set of landmarks in each of the images that are to be registered. The two images are then registered by relying on a known relationship among the landmarks in the two images. One limitation of this approach to image registration is that the registration accuracy depends on the number and location of landmarks selected. Selecting too few landmarks may result in an inaccurate registration. Selecting too many landmarks does not necessarily guarantee accurate registration, but it does increase the computational complexity of registration. Also, the manual operations required are time consuming. Furthermore, it is not always possible to identify appropriate structural landmarks in all images.
Recently, two different imaging modalities have been combined in a single imaging device. This integrated hardware approach to image registration is a less than optimal solution to the problem of image registration due to cost and logistical reasons. In many cases, hardware registration is impractical or impossible and one must rely on software-based registration techniques. For example, such a hardware approach is not applicable to the registration of images acquired at different times or from different subjects, e.g., when monitoring treatment effectiveness over time, or for applications involving inter-subject or atlas comparisons. Software registration would also be necessary in some cases, even if a hardware-based approach to registration is used. For example, software registration would be needed for the correction of motion that occurs between sequential scans taken on the same machine, such as transmission and emission scans in PET and SPECT, and for the positioning of patients with respect to previously determined treatment plans.
In recent years, full volume-based registration algorithms have become popular since they do not rely on data reduction, require no segmentation, and involve little or no user interaction. More importantly, they can be fully automated and provide quantitative assessment of registration results. Entropy-based algorithms, the mutual information approach in particular, are among the most prominent of the full volume-based registration algorithms. Most of these algorithms optimize some objective function that relates the image data from two modalities. These entropy or mutual information techniques are bounded in one direction, but unbounded in the other. For example, mutual information has a lower bound of 0, but the upper bound is implementation dependent, e.g., the number of bins. Likewise, entropy has a lower bound 0 and the upper bound is also implementation dependent. As a result, the degree to which the images are successfully registered is difficult to understand.
Accordingly, the present invention contemplates a new and improved image processing system and method which overcome the above-referenced problems and others.
SUMMARY OF THE INVENTION
In accordance with a first aspect, the present invention provides a method for registering first and second volumetric images comprising three-dimensional arrays of gray scale voxel values. Mutation probabilities are defined for a plurality of aligned pairs of voxel values comprising a voxel value from the first image and a spatially corresponding voxel value from the second image. The mutation probabilities can be obtained from previous statistics on registered images or calculated purely from the current data set. Each mutation probability is related to the likelihood that a voxel value in one image corresponds to a spatially corresponding voxel value in the other image and is based on a selected geometric relationship of the images. A first transform defining a geometric relationship of the second image relative to the first image is selected and a measure of the likelihood for a predetermined set of aligned voxel pairs using the mutation probabilities is calculated, the measure of the likelihood being an indicium of the probability of obtaining the first image given the second image and the probability of obtaining the second image given the first image. A different transform defining a geometric relationship of the second image relative to the first image is selected and the process is repeated in iterative fashion until an optimal transform providing an optimal measure of the likelihood is calculated.
In accordance with another aspect, an article of manufacture is provided comprising a computer useable medium having computer readable code means embodied therein for performing the method of the present invention.
In accordance with another aspect of the present invention, an image processing system for registering first and second volumetric images includes a registration processor and associated memory for storing a plurality of volumetric image representations to be registered, a memory coupled to the registration processor for storing paramet
Cochoff Steven M.
Zhu Yang-Ming
Au Amelia M.
Fay Sharpe Fagan Minnich & McKee LLP
Kim Chong
LandOfFree
Imaging registration system and method using likelihood... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Imaging registration system and method using likelihood..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Imaging registration system and method using likelihood... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3188300