Image analysis – Applications – Biomedical applications
Reexamination Certificate
1998-12-21
2001-07-31
Mancuso, Joseph (Department: 2621)
Image analysis
Applications
Biomedical applications
C128S922000
Reexamination Certificate
active
06269176
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to digital medical imaging, particularly the presentation of digital projection radiography images (acquired from a film digitizer, a computed radiography system or a flat panel direct digital detector) for improved image quality.
BACKGROUND OF THE INVENTION
The use of anti-scatter devices such as linear grids has been proven to improve contrast and signal to noise in radiographic images. Because of this, these devices are commonly found in most radiology departments and come in an assortment of configurations. A grid typically consists of a series of lead foil strips separated by xray-transmissive spacers. The spacing of the strips determines the grid frequency, and the height-to-distance between lead strips determines the grid ratio. Grids can be oriented horizontally or vertically. Two general methods of use exist for grids—stationary and moving (Bucky-Potter configuration). Of the moving type grids, the shadows of the lead strips are blurred out by the motion, which can be either reciprocating or unidirectional (single stroke). Conversely, stationary grids cause the shadows of the lead strips to be imposed onto the radiographic image. In cases where a moving grid is not reciprocating properly, or the time of exposure is faster than the time it takes for the grid to move, the resulting image will also have the lead strip shadows present.
Due to the migration of the radiology department from analog screen/film to digital imaging and PACS, issues once germane to the radiologist, such as the shadows of grid lines in the film images, are now becoming a source of displeasure and annoyance, particularly when primary diagnosis is being done on soft copy or a CRT. The cause of the problem is the high frequency of the lead strips found in the grids which cause artifacts in the image that resemble moire patterns and can hinder interpretation. The patterns are caused by aliasing which is introduced due to the discrete sampling of the image by the scanning system. Factors which contribute to the aliasing are the grid resolution (grid line frequency), the sampling frequency and the modulation transfer function (MTF) of the image acquisition device. The most typical manifestation of the problem occurs when an image is reduced in size for the purpose of soft copy display on a monitor/CRT at a slightly reduced size.
In order to develop a comprehensive solution to deal with the aliasing artifact, one must first detect if a stationary grid is present in the image. The use of stationary grids in most radiography departments, although not random, is sporadic. And because the most commonly used grid configuration, the moving Bucky-Potter, can experience malfunction, prediction based upon exam information can be erroneous. Therefore, an image analysis technique is required to automatically detect the presence or absence of a grid in every image. U.S. Pat. No. 5,661,818, issued Aug. 26, 1997, inventors Gaborski et. al., developed a grid detection method which bases its detection decision on a double auto-correlation calculation. Variances are measured independently, both horizontally and vertically and a statistical F test is performed to determine if the variances are the same over a randomly chosen sampling of locations within the image. Votes are then tallied and if a majority exists that the variances are different, a decision is made in favor of a grid being present. However, this method is limited because it does not provide any characteristic information about the nature of the grid if one is detected—the grid line frequency(s), the noise power of the grid, etc. Such information is critical to the second stage of an automated solution to deal with grid aliasing—that of suppressing the grid lines.
Once a grid is detected, the presence of the grid shadows needs to be either removed or suppressed. These shadows can be considered a form of noise and in the field of digital image processing, this type of noise artifact is referred to as correlated noise. Well known methods exist to characterize and eliminate correlated noise (see “Noise Cleaning”, W. K. Pratt,
Digital Image Processing
, 2
nd
Edition, John Wiley & Sons, Inc., 1991). However, the frequency of grid lines within a given device are quite variable. This is due to the manual nature of the manufacturing process. Because of this, 2-D Fourier filtering methods involving the use of a bandstop filters are less straight forward and are prone to the introduction of artifacts if the filter is incorrectly designed. Also, commercial viability of such methods requires special purpose hardware due to the relatively large format of the image (2K×2K up to 4K×4K, 12 bits/pixel) in order to meet near real-time speed requirements. Spatial filtering is the next best choice, such as convolution with a blurring filter. But, such a solution results in a global reduction of image detail if applied indiscriminately. Adaptive filtering methods are therefore appropriate.
U.S. Pat. No. 4,792,900, issued 1988, inventors Sones et al., describe a method of adaptive noise suppression used in dual energy digital radiography. Both low and high energy images are filtered in two dimensions, based upon the pixel value of the respective images using either bone specific or soft tissue specific filter functions which are of a predefined size (filter support not dependent upon noise being suppressed). Although the method taught by Sones, et. al. does somewhat utilize custom filter functions as a function of signal level, the method is not designing the filters as a function of the noise which it is trying to suppress, nor are the filters gaussian in shape. Not adapting the filter size on an image by image basis in order to meet the noise blurring objective is a significant deficiency, as it is in U.S. Pat. No. 5,764,307, issued 1998, inventors Ozcelik, et. al. Ozcelik et. al. describe a method for spatially adaptive filtering for video encoding where the displaced frame difference (DFD), i.e. the temporal difference between 2 consecutive frames, is blurred to reduce compression artifacts. This approach also uses a 2-D fixed filter support (size). In the present invention, such 2-D filtering methods would be too indiscriminate because of the 1-D orientation of the grid shadows. Thus the present invention enjoys the benefits of custom filter design on an image-by-image basis, faster processing and the preservation of detail in the orientation corresponding to the orientation of the grid.
In accordance with the present invention, there is provided a new and comprehensive image processing system which overcomes the above referenced problems.
SUMMARY OF THE INVENTION
The present invention provides an automatic method to quickly detect and suppress grid lines in projection radiographic images. More specifically, the grid detection method includes a means for detecting if a grid is present, and if present, determining what the characteristics of the grid are in terms of the frequency of the grid, the magnitude (in terms of 1D Fourier spectra) of the grid peak, the spread of the grid peak (in terms of the half-width of full maximum of the grid peak), the total energy of the grid peak, the orientation of the grid, and a figure-of-merit assessing the confidence of the detection. The grid suppression method includes a means for building a series of one dimensional gaussian blur filters based upon the grid characteristics and a means for adaptively applying the filters. A plurality of methods to adaptively apply the filters is described.
ADVANTAGEOUS EFFECT OF THE INVENTION
The present invention has the following advantages:
1. The combined grid detection and suppression methods solve the problem of moire artifacts caused by grid aliasing in digital projection radiographic images. No other prior art was found which does so.
2. The grid detection method both determines if a grid is used, and if so, what the characteristics of the grid are such as grid frequency, magnitude, energy, the spread of the grid peak, the orientation
Barski Lori L.
Wang Xiao-hui
Eastman Kodak Company
Mancuso Joseph
Noval William F.
Tabatabai Abolfazl
LandOfFree
Method for x-ray antiscatter grid detection and suppression... 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 x-ray antiscatter grid detection and suppression..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for x-ray antiscatter grid detection and suppression... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2554846