Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
1998-04-17
2001-01-30
Couso, Yon J. (Department: 2723)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S275000
Reexamination Certificate
active
06181832
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to electronic imaging and image registration, and more particularly, to the reduction of noise introduced by post-processing of electronic images.
BACKGROUND OF THE INVENTION
Images such as photographs, x-rays, and maps contain an abundance of information. Often such images are recorded as electronic images to allow digital storage and indexing, rapid access, and post-processing including constructing a composite image from multiple existing images, filtering existing images, and removing noise from existing or constructed images.
Magnetic resonance imaging (MRI) is a widely-used diagnostic technique in which radio frequency (RF) signals are analyzed to produce diagnostic information and stored as electronic images. Echo-planar imaging (EPI) is a subset of MRI that provides high temporal resolution achieved with faster imaging techniques, and typically results in large image data sets. One application of EPI is functional magnetic resonance imaging (fMRI), where a time series of images are acquired for a selected plane (or planes) of a subject. In challenge-based fMRI of the human brain, a time series of images of one or more planes within a subject's brain are collected while the subject is exposed to a sequence of stimulus conditions, to identify functional changes in brain characteristics.
The high spatial resolution of EPI makes challenge-based experiments sensitive to subject movements on the scale of millimeters or less. Stimulus-correlated movements of the subject may lead to false results. The false information introduced by a movement by the subject is commonly referred to as a motion artifact. To accurately analyze the time series of images, motion artifacts must be removed by registering the series of images. Proper interframe registration of the time series of images to remove motion artifacts is particularly important in studies in that the subject's motion is an integral part of the experiment, such as experiments which require spoken responses or the performance of motor tasks. See, for example, J. V. Hajnal et al.,
Magn. Reson. Med.,
31:283-291, 1994.
Techniques used to register image sets include the use of physical immobilization devices to maintain the subject in a known position and the placement of external fiduciary markers as landmarks for subsequent alignment. There are also numerous registration algorithms that construct “registered” images by repositioning each of the “misregistered” images by linear interpolation or Fourier regridding. See, for example, L. C. Maas et al.,
Magn. Reson. Med.,
37:131-139, 1997, and A. Apicella et al.,
SPIE Medical Imaging III: Image Processing,
1092:252-263, 1989.
However, such registration algorithms can introduce additional artifacts into the image set that corrupt fMRI analysis.
SUMMARY OF THE INVENTION
The invention features methods and systems for using and determining post-registration correction terms to remove resampling artifacts from a series of registered images. The invention is based on the discovery that the artifacts introduced into a series of registered images by a registration algorithm generate variances at multiple locations of the images, e.g., at each point of the images, that can be modelled by an incomplete set of basis functions, which are referred to as post-registration correction terms. For example, post-registration terms including at least one non-linear polynomial, e.g., the first six Legendre polynomials, are useful for modelling resampling artifacts produced by rotational registration. The rotational angles used by the rotational registration are used as the arguments in the polynomial post-registration correction terms.
To remove resampling artifacts from registered images, a processing system determines the variance between the series of registered images, at each of multiple locations, that can be expressed in terms of a superposition of the post-registration correction terms, e.g., by a multiple linear regression. The processing system then removes that variance from the series of registered images at each of the multiple locations.
In general, in one aspect, the invention features a method for determining a set of post-registration correction terms that correct for artifacts introduced by an algorithm used to register images represented by values at a first array of points. The method includes: providing a series of images of a reference object, wherein each image in the series corresponds to a particular orientation of the reference object; registering the images with each other using the algorithm to produce a series of registered images in which the reference object in each of the registered images is in a common orientation, wherein each of the registered images is represented by registered values at a second array of points; and determining the set of post-registration correction terms based on variations among the registered values of the registered images at selected points of the second array.
The method can include any of the following features. Each post-registration correction term can be a vector having a factor for each of the particular orientations of the reference object. Alternatively, each of the post-registration terms can be a function that depends on a coordinate corresponding to the particular orientations of the reference object. The first and second arrays can be two-dimensional or three-dimensional. The particular orientations of the reference object can be along a rotational coordinate or a translational coordinate. The first and second arrays can be identical to one another. The method can determine the set of post-registration correction terms from principle component analysis of the variations among the registered values of the registered images at the selected points of the second array.
In another aspect, the invention features a method for correcting artifacts introduced by registering images with an algorithm, wherein the registered images are represented by values at an array of points. The method includes: determining a set of post-registration correction terms for the algorithm according to the method for determining a set of post-registration correction terms described above; and comparing a variation among the values of the registered images at a selected region of the array with a variation expressed by a superposition of the post-registration correction terms.
The method for correcting artifacts can include any of the following features. Comparing can include performing a regression on the values of the registered images at the selected region of the array in terms of the post-registration correction terms, and if desired, one or more stimulus terms. Comparing can also include comparing a variation among the values of the registered images at an additional selected region of the array with a variation expressed by a superposition of the post-registration correction terms. The selected region of the array can be a selected point of the array. Comparing can also include comparing variations among the values of the registered images at each point of the array with a variation expressed by a superposition of the post-registration correction terms.
Comparing can also include performing a regression on the values of the registered images at selected regions of the array in terms of the post-registration correction terms and a stimulus term to determine a regression value for each of the post-registration correction terms and the stimulus term at each of the selected regions. In this case, the method can further include constructing an output image based on the regression value for the stimulus term at each of the selected regions, or the stimulus term and a regression residual for each of the selected regions.
The method can also include subtracting from the values of the registered images at the selected region of the array the superposition of the post-registration correction terms.
In another aspect, the invention features a second method for correcting artifacts introduced into images
Couso Yon J.
Fish & Richardson P.C.
McLean Hospital Corporation
LandOfFree
Methods and systems for removing artifacts introduced by... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and systems for removing artifacts introduced by..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and systems for removing artifacts introduced by... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2443820