X-ray or gamma ray systems or devices – Electronic circuit – With display or signaling
Reexamination Certificate
1998-07-11
2001-01-16
Church, Craig E. (Department: 2876)
X-ray or gamma ray systems or devices
Electronic circuit
With display or signaling
C378S098200
Reexamination Certificate
active
06175613
ABSTRACT:
The invention relates to the processing of a sequence of radiological images of an object.
After an object, for example a part of a patient's body, has been illuminated with X-radiation, a radiological image of the object is obtained. In practice, for the purpose of obtaining the corresponding radiological image, the illumination of the object gives rise to direct radiation and radiation which is Scattered by the object itself. However, this scattered radiation leads to the addition of unsharpness on the radiological image which is obtained, and this makes it more difficult to pick out elements of interest on the acquired radiological image, for example specific parts of the human body which are intended to be examined.
In general, in the processing of a sequence of radiological images of an object, after having acquired the raw image its scattered radiation or “unsharpness” is estimated and at least partially subtracted from the total image.
Several solutions have to date been proposed for estimating the scattered radiation of a radiological image.
One solution may consist in measuring the scattered radiation at various locations of the image while masking the primary radiation, for example with one or more discs, and by then interpolating this measurement over the total image. One drawback of this solution resides in the loss of a useful fraction of the information due to the presence of “holes” in the image, these holes being caused by the masking discs.
Another solution consists in acquiring two images of a patient while using different anti-scatter screens, then by using these two acquired images to reconstruct an image which is free of the scattered radiation. However, a solution of this type has the drawback that an additional image is acquired, which increases the patient's exposure time to the X-radiation.
All these solutions therefore involve additional physical means which are more or less complex and expensive.
Another solution consists in estimating the level of the scattered radiation on the basis of the parameters of the acquisition, in particular while taking the dimensions of the object into account. However, a solution of this type does not make it possible for the internal content of the object to be taken into account.
Theoretical methods have therefore been proposed for estimating the scattered radiation on the basis of the acquired image without using additional physical means. According to this theory, the scattered radiation is proportional to an average convolution weighted by a two-dimensional exponential mathematical function (kernel) taken over a moving window. In other words, the scattered radiation is estimated on the basis of a low-pass filter of the image, the ideal impulse response of which should have rotational symmetry, a decreasing exponential shape and a parameterizable full width at half-height.
Nevertheless, implementation of this theory leads to significant complexity of the corresponding algorithm, as well as to long computation times. Furthermore, for each pixel of the image, and in particular for the first pixel, implementation of this wide exponential function entails acquiring all the pixels of the image before the processing can be carried out. The result of this is that the processing has a latency time which may be as long as 30 ms. However, the time between the acquisition of two successive images is generally of the order of 33 ms. Further to the latency caused by the processing, the other phases of the processing, namely the acquisition and display in particular, themselves lead to latency times of non-negligible length, for example 25 ms for the acquisition. In consequence, further to the problems of algorithm complexity and computation time, implementation of a low-pass filter of this type may lead to total latency times of 100 ms, and this may prove problematic for examining the images, in particular when a catheter present in the patient's body moves.
The invention aims to provide a solution to these problems.
One object of the invention is to estimate the scattered radiation on the basis of the acquired image without using additional physical means, and to propose a very simple way of implementing the estimate of the scattered radiation on the basis of an average convolution weighted by a rotationally symmetric decreasing exponential function, taken over a moving window.
According to the invention, this implementation method greatly reduces the computation time and the algorithm involved, and minimizes the latency time of the processing, making it compatible with a typical acquisition rate of radiological images.
The invention therefore proposes a method for processing a sequence of radiological images or an object, comprising a step of estimating, for each current image, the radiation scattered by the object, and an image correction step in which an estimated scattered radiation is at least partially eliminated from the current image.
According to a general characteristic of the invention, in the image correction step, the scattered radiation estimated and at least partially eliminated from the current image is the scattered radiation estimated for the previous image. In combination with this characteristic, the estimation step provides the definition of a first row scan direction for the pixels (for example from left to right) and a second row scan direction, which is the opposite of the first, (for example from right to left), for each row of the current image, and a first column scan direction for the pixels (for example top-down) and a second column scan direction, which is the opposite of the first, (for example bottom-up), for each column of the current image. A recursive law is furthermore defined which, for a pixel in question, develops a so-called calculated intensity, this calculated intensity being obtained by modulating the calculated intensity of the previous pixel, while taking into account the scan direction in question, with a coefficient of less than 1, and by adding to this modulated intensity a so-called initial intensity of the said pixel in question, modulated with the complement to one of the said coefficient.
Advantageously, during the first application of the said recursive law using one of the scan directions, for example in the left to right direction, for a row, the said initial intensity of the current pixel is the intensity of the pixel in the current acquired image, whereas for the subsequent applications, the initial intensity of the current pixel is that obtained from the previous applications of the said recursive law.
According to the invention, in the estimation step, the said recursive law is applied four times in succession to each current pixel of the current image, while respectively considering the two row scan directions and the two column scan directions. The intensity calculated for the said current pixel after the four applications of the recursive law is then representative of the value of the scattered radiation for this pixel.
This improves the quality of the images while increasing the image processing speed and minimizing the latency time of the processing.
According to a preferred embodiment of the invention, for each current row of the current image, the said law is applied a first time using the first row scan direction (for example from left to right), for all the pixels of the row, then the said law is applied to the said current row for a second time using the second row scan direction, (that is to say in the opposite direction starting from the last pixel of the row), for all the pixels of the row, then, when all the pixels of the said current row have been taken into consideration using the said first and second row scan directions, the said recursive law is applied a third time for each current pixel of the current row by using this current pixel and the pixel which is located in the same column and precedes this current pixel in terms of the first column scan direction. Thus, for example, the current pixel and the pixel located above this current
Boutenko Vladislav
Klausz R{acute over (e)}my Andre
Vaillant R{acute over (e)}gis
Chaskin Jay L.
Church Craig E.
GE Medical Systems S.A.
LandOfFree
Method for processing a sequence of radiological images of... 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 for processing a sequence of radiological images of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for processing a sequence of radiological images of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2478030