Surgery – Diagnostic testing – Detecting nuclear – electromagnetic – or ultrasonic radiation
Reexamination Certificate
2001-08-10
2003-08-12
Jaworski, Francis J. (Department: 3737)
Surgery
Diagnostic testing
Detecting nuclear, electromagnetic, or ultrasonic radiation
Reexamination Certificate
active
06605042
ABSTRACT:
BACKGROUND OF INVENTION
Certain embodiments of the present invention generally relate to ultrasound imaging for the purpose of medical diagnosis. In particular, certain embodiments of the present invention relate to calculating the rotation angle and the linear shifts between successive ultrasound image frames in order to obtain an extended field of view ultrasound image.
Ultrasound systems are commonly used to produce images of internal anatomy. An ultrasound system may, for example, be used to view the anatomy of a human patient for the purpose of diagnosing disease or determining an appropriate course of medical treatment. The system utilizes an ultrasound probe to acquire an image of the patient's anatomy. The size of the field of view of the acquired image is limited by the aperture size of the probe. Unfortunately, the region of interest within the patient may be larger than the available field of view. In this example, the probe is moved to acquire individual image frames of the region of interest, where such frames were outside the probe's initial field of view. Viewing large regions of interest in this manner, wherein only a portion of the anatomy of interest can be viewed at one time, may be confusing and could potentially limit the ability of the operator to correctly diagnose the patient.
To overcome the size limitation of the available field of view, the imaging technique referred to as “Extended Field of View” (EFOV) may be used. EFOV imaging is useful in various clinical applications including comparisons of adjacent anatomical structures, and in viewing enlarged organs and large abdominal masses.
In EFOV imaging, the probe is moved smoothly and continuously over the desired region of interest. As the probe is moved, the ultrasound system acquires a series of consecutive images. The consecutive images may be adjacent to each other, or the ultrasound system may discard some images and retain other images for processing. For example, one image may be retained for every three images acquired. The retained images are combined by the ultrasound system to form an image of the entire region of interest. The diagnostic value of the EFOV technique largely depends upon the accuracy of the dimensions and relative locations of organ structures in the EFOV image, thus the series of consecutive images must be correctly related to each other. In other words, the images must be adjusted linearly, in one or both of the x and y directions, and rotated as necessary to form an accurate compound image. This process is known as image registration.
The image registration in EFOV imaging computes relative linear shifts between image frames. Because the patient surface is curved in most situations, the image registration method should also accurately estimate the rotational shift (or angle) between a series of successive images. Various registration methods are used in EFOV imaging, including sum of absolute difference (SAD), landmark matching, pattern recognition and analysis, and the like. Among these, the SAD method has been found to be more effective and less computationally intensive than the other registration methods.
Previously, the SAD method has been used to compute both the linear shifts and the rotation angle between consecutive image frames in a series. To compute the linear shifts, the SAD method divides two consecutive image frames into kernels, and then compares the pixel values of the kernels of one image frame to the pixel values of the corresponding kernels of the other image frame. The kernels of one image are moved within a defined boundary to find the location within the boundary that is the most similar to the corresponding kernel of the other image. In a similar manner, to compute the rotation angle, the SAD method rotates one image frame through a predetermined range of degrees at a preset step size. Each time the image frame is rotated, the pixel values are compared to the pixel values of the second image frame. The angle at which the two image frames are the most similar is determined to be the angle of rotation between the two image frames.
Although the SAD method is effective in computing linear shifts, it is very time consuming when used to compute the rotation angle. In addition, the SAD method utilizes a preset discrete step size when rotating one image frame in relation to the other image frame. To reduce computation time, the step size may be set somewhat course. Unfortunately, it is possible that the most accurate rotation angle is an angle that occurs between two preset steps. Decreasing the preset step size, however, would further increase the time required to compute the rotation angle, and is thus not desirable. Because EFOV images may typically cover large angles, such as 90 degrees, and may be comprised of up to 1000 individual image frames, any increase in computation time is not desirable. Thus, a need has long existed in the industry for a method for calculating rotation and linear image registration in EFOV imaging that addresses the problems noted above and other problems that will become apparent from the following description.
SUMMARY OF INVENTION
In accordance with at least one embodiment, a method is provided for obtaining an extended field of view diagnostic ultrasound image. First and second image frames of an object of interest are acquired. The image frames are rotated relative to one another. Pixel points, representing spatial points in the object of interest, are identified in the first image frame. Pixel points corresponding to the pixel points in the first image frame are then computed in the second image frame. A rotation angle between the first and the second image frames is calculated based on a least-squares relation between the pixel points. The first and second image frames are combined to form an extended field of view image.
In accordance with at least one embodiment, multiple registration kernels are identified in the first and second image frames, and a search region containing a kernel is identified in the second image frame. The pixel points are identified by a set of coordinates. In one embodiment, the rotation angle is calculated using the coordinates of the pixel points and a number representing the number of registration kernels in each of the image frames. In an alternative embodiment, the image frames are acquired successively in time.
In accordance with at least one embodiment, the image frames are linearly shifted relative to one another. The linear shift is calculated based on a sum of absolute differences between the pixel points in the image frames. In an alternative embodiment, the linear shift is calculated based on a least-squares method.
In accordance with at least one embodiment, a method is provided for computing the image registration of two images. Consecutive image frames are acquired. The image frames represent partially overlapping portions of a region of interest. At least one pixel point in each image frame representing one spatial point in the region of interest is identified. In an alternative embodiment, the pixel points of the second image frame are computed based on a sum of absolute differences. The squared difference between the pixel points in consecutive image frames is calculated to obtain a rotation angle between consecutive image frames. In an alternative embodiment, the squared difference comprises a least-squares relation between the pixel points. In another embodiment, the rotation angle is derived from a continuous range of possible rotation angles.
In accordance with at least one embodiment, multiple registration kernels are identified in each image frame. The squared difference is calculated by using a value representing the number of registration kernels located in each image frame and the pixel coordinates. In an alternative embodiment, the first and second coordinates are summed over a predetermined range to calculate the squared difference.
In accordance with at least one embodiment, a linear shift between the consecutive image frames is calculated based on the
Dong Fang F.
Geiser Brian Peter
Dellapenna Michael A.
GE Medical Systems Global Technology Company LLC
Jaworski Francis J.
McAndrews Held & Malloy Ltd.
Patel Maulin
LandOfFree
Method and apparatus for rotation registration of extended... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for rotation registration of extended..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for rotation registration of extended... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3127912