Surgery – Diagnostic testing – Detecting nuclear – electromagnetic – or ultrasonic radiation
Reexamination Certificate
1998-03-13
2001-02-20
Lateef, Marvin M. (Department: 3737)
Surgery
Diagnostic testing
Detecting nuclear, electromagnetic, or ultrasonic radiation
C324S309000
Reexamination Certificate
active
06192265
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a diagnostic image processing method in which a series of two-dimensional images is formed of an object present in an examination zone, each image of the series being derived from a number of one-dimensional projection images which are formed by projection of the object from one of a plurality of predetermined projection directions and which consist of pixels.
In the MR (magnetic resonance) technique methods of this kind are known in conjunction with the so-called projection-reconstruction method. Such methods are also used in X-ray computer tomography.
2. Description of Related Art
For medical diagnosis it is often necessary to form a series of images of one and the same object, said series representing the object in, for example different phases of motion. The quality of the individual images is then usually limited by noise. In the case of MR methods this noise occurs if the MR signals required for the reconstruction of the images are acquired in an as short as possible period of time. In the case of X-ray CT methods noise occurs whenever only a small dose is used per individual image so as to keep the radiation dose for the patient small.
An article by T.A. Reinen in Proc. SPIE, Vol. 1606, pp. 755-763, 1991, discloses a method which reduces the noise in such a series of images by determining a motion vector field, by applying a noise reduction filter to the pixels (pixel=picture element) linked by the motion vector field, and by taking up the image values resulting from the filtering operation in an output image. This method is based on the consideration that even though the individual images of the series deviate from one another because at least some details are situated in different positions within the images, these details are usually reproduced with the same brightness in the various images. This makes it possible to link two images of the series by way of a motion vector field which describes the shift of these details from one image to another and hence links the content-wise associated pixels (i.e. pixels reproducing the same point of a structure in the various images). The known noise reduction method could also be applied to series of images formed by means of the above-mentioned methods.
It is a drawback of this method, however, that it is comparatively time consuming because for each pixel in an image in prince the content-wise associated pixel within a two-dimensional image window in the other image must be searched. Citation of a reference herein, or throughout this specification, is not to construed as an admission that such reference is prior art to the Applicant's invention of the invention subsequently claimed.
SUMMARY OF THE INVENTION
Therefore, it is an object of the present invention to conceive an image processing method of the kind set forth in such a manner that the noise reduction can be performed faster than by means of the known noise reduction method.
This object is achieved according to the invention in that the following steps are carried out:
a) determining a motion vector field which links the content-wise associated pixels in projection images with the same and/or the opposite projection direction,
b) applying a noise reduction filter to the image values of the pixels linked by the motion vector field and possibly to their neighboring pixels,
c) taking up the image values resulting from the filtering operation in an output projection image,
d) reconstructing the images from the associated output projection images.
Thus, according to the invention the noise reduction method is not applied to the individual pixels of a two-dimensional MR or CT image, but to the pixels of a one-dimensional projection image; this offers the advantage that content-wise associated pixels must be searched only within a one-dimensional window. Therefore, this method is faster. The two-dimensional MR image or CT image is derived from a number of one-dimensional projection images which have been noise-filtered in this manner.
One possibility of determining a motion vector field for a projection image includes the following steps: selecting a block (b) of associated pixels in this projection image (B
i
), determining the position of a block which is present in at least one of the preceding or succeeding projection images (B
i−1
), produced with the same and/or the opposite projection direction, said block having the best correspondence with the block (b) in the one projection image, deriving the motion vector from the position of the block or blocks (b′) thus determined, and repeating the previous two steps for other blocks in the projection image (B
i
). The blocks showing the best similarity can be determined by applying (
203
) a similarity measure in order to evaluate the correspondence of a block (b) in the one projection image (B
i
) with a block (b′) which is present in different positions in the other projection image (B
i−1
), and determining (
204
) the position for which the optimum similarity measure is obtained. The determination of the position with the optimum similarity measure as carried out therein can be performed by determining the similarity measure for a starting position of the block (b′), changing (
205
) the position of the block (b′) and determining the similarity measure for the changed position, deriving a new position from the similarity measures for the preceding positions and determining the similarity measure for this position, and iterative repetition of the previous step.
Another possibility of determining a motion vector field is characterized in that it includes the following steps: forming (
301
,
302
) at least one projection image (B
1i
, B
2i−1
), having a reduced spatial resolution and a reduced number of pixels, for each projection image of the series, determining (
303
) motion vectors which link the content-wise associated pixels in different projection images (B
2i
, B
2i−1
) of the series, and repeating (
304
) the previous step for the projection images (B
1i
, B
1i−1
) with the next higher spatial resolution using the previously determined motion vectors. The determination of the motion vector field first with a comparatively coarse and subsequently with an increasingly finer resolution, yields fast determination of the motion vector field which is insusceptible to artefacts to a high degree.
A preferred noise filter includes a weighted summing of the image values of the pixels which are content-wise associated with, or are neighbors of this pixel in the other projection images, the weighting factors being greater as the difference between the images value of this pixel and that of one of said pixels is smaller.
A method which is focused on one-dimensional projection images includes the determination, involving a plurality of cycles, of a motion vector for a pixel (j) of a projection image, each cycle including the following steps: selecting different motion vectors (t, s, m), at least one of which can be varied from one cycle to another, evaluating, using a similarity measure (M
1
. . . M
4
), the blocks associated with one another by each motion vector (m), selecting the motion vector (m) having the most attractive similarity measure, and repeating the previous three steps during the next cycle. A device suitable for carrying out the method according to the invention includes means for forming a series of images of an object present in an examination zone, each image of the series being derived from a number of projection images which are formed by projection of the object from one of a plurality of predetermined projection directions and which consist of pixels, and a programmable image processing unit (
24
) for processing the projection images, characterized in that the image processing unit (
24
) is programmed in such a manner that the following steps are executed: determining (
101
,
102
) a motion vector field which links the content-wise associated pixels in proje
Carlsen Ingwer
Grass Michael
Rasche Volker
Van Vaals Johannes J.
Lateef Marvin M.
Mantis Mercader Eleni
U.S. Philips Corporation
Vodopia John F.
LandOfFree
Diagnostic image processing method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Diagnostic image processing method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Diagnostic image processing method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2575492