Surgery – Diagnostic testing – Detecting nuclear – electromagnetic – or ultrasonic radiation
Reexamination Certificate
2000-02-03
2002-10-29
Lateef, Marvin M. (Department: 3737)
Surgery
Diagnostic testing
Detecting nuclear, electromagnetic, or ultrasonic radiation
C250S363020, C382S128000, C128S922000
Reexamination Certificate
active
06473636
ABSTRACT:
BACKGROUND
1. Technical Field
The present invention is generally related to medical imaging and, in particular, to a variable-length method for correcting body motions and heart upward creep in Single-Photon Emission Computed Tomograph (SPECT) myocardial perfusion imaging in nuclear medicine.
2. Background
The single-photon emission computed tomograph (SPECT) myocardial perfusion imaging has been widely used in the diagnosis of coronary artery diseases. Since the whole image acquisition process usually takes more than 20 minutes, it is difficult for the patient to completely lie still during the acquisition. Also, heart upward creep occurs frequently during stress studies. Both heart upward creep and patient body motion during myocardial SPECT data acquisition may cause image artifacts (distortions of the appearance of the object of interest, such as the misalignment of the anterior and posterior walls of the left ventricle), resulting in potential false diagnosis. To reduce false positive diagnosis, both body motions and heart creep need to be post-corrected.
Prior art methods for correcting these motions are generally restricted to abrupt body motions. A brief description of some of these prior art methods will now be given.
The decomposition of body motion into axial (y) and trans-axial (x) components by projecting the images onto the x- and y-axes of the image plane is described by Eisner et al., in “Use of Cross-Correlation Function to Detect Patient Motion During SPECT Imaging”, Journal of Nuclear Medicine, Vol.28, No.1, 1987, pp.97-101. Each motion component is estimated independently by cross-correlating the projected images of adjacent frames. However, as noted in the Eisner article, this procedure is unable to detect both gradual body motion and heart creep, since adjacent image frames for gradual motions and heart creep do not exhibit significant differences. Moreover, because of information loss in the projection of a 2D image into two 1D images, the method may either underestimate or overestimate the motions.
U.S. Pat. No. 5,758,645, entitled “Automatic Detection and Correction of Body Organ Motion and Particular Cardiac Motion in Nuclear Medicine Studies”, issued on Jun. 2, 1998, assigned to the assignee herein, the disclosure of which is incorporated herein by reference, describes the estimation of body motions by minimizing the second order derivatives of the 1D projection images. However, this method does not explicitly address gradual body motions and heart creep.
The diverging square method, described by Geckle et. al., in “Correction for Patient and Organ Movement in SPECT: Application to Exercise Thallium-201 Cardiac Imaging”, Journal of Nuclear Medicine, Vol.29, No. 4, 1998, pp.441-450, uses a growing square to track the heart. If the heart could be tracked accurately, virtually all kinds of motions could be dealt with. Unfortunately, for rest-injected and redistribution Thallium studies with high uptake regions such as the liver, the square may be wrongly trapped. Also, since hearts vary in size, the use of a fixed upper limit for growing the square may cause errors.
The two-dimensional fit method, described by Cooper et. al., in “Detection of Patient Motion During Tomographic Myocardial Perfusion Imaging”, Journal of Nuclear Medicine, Vol.34, No.8, 1993, pp.1341-1348, puts a circular “Region Of Interest” (ROI) on the heart, and finds the body motion by correlating the ROI of the current image with the next image. This method was reported to be able to more accurately estimate motions than the decomposition-based 1D method. However, similar to the decomposition-based 1D method, gradual body motions and heart upward creep were not addressed.
In the prior art approaches, an ROI usually needs to be selected manually to cover the heart and to turn off effects of non-heart motions, such as bowel gas motions. This induces interactive work. Accordingly, it would be desirable and highly advantageous to have a fully automatic method for accomplishing the same. In particular, it would be desirable and highly advantageous to have an automatic method for correcting both body motion and heart upward creep in SPECT myocardial perfusion imaging.
SUMMARY OF THE INVENTION
The present Invention is directed to a method for motion correction in SPECT myocardial perfusion imaging. The method detects and corrects abrupt body motions, gradual body motions, and heart upward creep.
According to a first aspect of the invention, a variable length correlation method is provided for compensating for body motions and heart creep in a sequence of image frames obtained by single-photon emission computed tomograph (SPECT) myocardial perfusion imaging. The method includes the step of determining whether a bright spot corresponding to a non-heart motion exists in a particular image frame. The y-coordinate lower limit of the particular image frame is set as an offset from the position of the bright spot, when the bright spot exists. The y-coordinate lower limit of the particular image is set as an offset from the bottom of the particular image frame, when the bright spot does not exist. Abrupt body motions, gradual body motions, and heart upward creep are detected and corrected in the sequence of image frames, based on varying correlation lengths. The detecting and correcting step is applied in the two-dimensional image domain.
According to a second aspect of the invention, the determining step includes the step of determining consistency of sizes and positions of binarized image blobs using multiple thresholds.
According to a third aspect of the invention, the determining step includes the step of binarizing a given image frame by successively lowering a threshold value. Connected regions in the given image are identified. Pre-identified characteristics of each of the connected regions are determined. A given connected region is identified as a bright spot corresponding to a non-heart motion, when the pre-identified characteristics of the given connected region correspond to predetermined criteria.
According to a fourth aspect of the invention, the pre-identified characteristics include at least some of the area, the center of gravity, the x-size, and the y-size.
According to a fifth aspect of the invention, the predetermined criteria include at least some of the x and y size-ratio being within a first pre-specified limit, the area being within a second pre-specified limit, and continuous changes across the multiple thresholds.
According to a sixth aspect of the invention, the detecting and correcting step includes the step of comparing a given image frame to earlier image frames of varying distances from the given image frame.
According to a seventh aspect of the invention, the detecting and correcting step includes the step of comparing a given image frame to at least one earlier image frame having a predefined distance from the given image frame, for each of abrupt body motions, gradual body motions, and heart upward creep, respectively.
According to an eighth aspect of the invention, abrupt body motions, gradual body motions, and heart upward creep are each determined and corrected during a separate pass of the method. Each of the separate passes include the step of comparing a given image frame to at least one earlier image frame having a predefined distance from the given image frame.
According to a ninth aspect of the invention, a first pass detects abrupt body motions, a second pass detects gradual body motions, and a third pass detects heart upward creep.
According to a tenth aspect of the invention, the detecting and correcting step includes the step of applying a linear transformation to at least some of the image frames to compensate for the x-directional appearance changes caused by viewing angle differences.
According to an eleventh aspect of the invention, the applying step includes the step of applying the linear transformation to one of every two of the image frames.
These and other aspects, features and advantages of the present invention will become apparent from the f
Chen Eric Q.
Engdahl John C.
Qian Jianzhong
Wei Guo-Qing
Mantis Mercader Eleni
Paschburg Donald B.
Siemens Corporate Research Inc.
LandOfFree
Variable-length correlation method for motion correction in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Variable-length correlation method for motion correction in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Variable-length correlation method for motion correction in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2998857