Image analysis – Image enhancement or restoration
Reexamination Certificate
1998-10-14
2002-09-03
Rogers, Scott (Department: 2624)
Image analysis
Image enhancement or restoration
C382S128000, C382S275000, C382S260000
Reexamination Certificate
active
06445830
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to the field of discrete pixel imaging systems, such as magnetic resonance imaging systems. More particularly, the invention relates to a technique for correcting or adjusting portions of image data to compensate for effects of image data filtration and expand to the field of view in such images.
BACKGROUND OF THE INVENTION
A variety of systems are known and presently in use for producing discrete pixel images. Such systems generally collect raw image data from a subject of interest over a desired imaging area, and convert the collected raw image data to data which can be used to reconstruct an image. In magnetic resonance imaging systems, for example, spins of gyromagnetic material in a subject of interest are altered and spatially encoded to produce emissions which are detected during a data acquisition phase of imaging. The sensed signals are filtered and further processed to produce digital data for each individual volume element or voxel in a desired slice of the subject. The voxel data is then processed to produce a corresponding set of discrete picture element or a pixel data which can be used to reconstruct a two-dimensional image. The resulting images may be used by radiologists, attending physicians, and clinicians to diagnose and treat physical conditions of patients. Similar processes are also employed outside the medical imaging field in an increasing number of applications.
Features of interest, such as anatomical features, are imaged in magnetic resonance imaging systems by displaying the frequency content of received echo signals from the gyromagnetic material across a display area or image matrix. The position of the pixel data within the image is directly related to the frequency of the sensed signal. This correlation is generally known in the art as frequency encoding. The unique frequency for each position, i.e., the frequency encoding, is created by varying gradient magnetic fields produced by gradient coils in the imaging system. While the emissions from the subject are sensed in real time, the resulting time domain data must therefore be transformed to the frequency domain to assign to the data the proper location in the subject. This transformation is performed following band limit filtering on the raw image signals.
A number of image quality parameters typically interact to define band limit filtering requirements. Such parameters include alias content, brightness or intensity uniformity, and signal-to-noise ratio. Aliasing, for example, tends to replace energy into the discrete band bins used to digitally reconstruct the image, and can thus result in image artifacts, such as “ghosting”. Alias content may be adjusted by moving a low pass filter limit or corner to sufficiently low frequencies to eliminate the undesirable aliasing. Because continuous domain or analog filters are employed in the image processing the corner of the low pass filter limit is not sharp, but slopes downwardly near the limit of the useful frequency range. A transition region is therefore established at the higher frequencies, i.e. beyond the corner transition of the filter. In typical systems, data in portions of the frequency band beyond the limit of the low pass filter, found in the transition region, is discarded in a process commonly referred to as stop band rejection. While this practice reduces the incidence of image artifacts, it also reduces the field of view of the image due to the relationship between frequencies of the signals and positions in the image.
Magnetic resonance imaging systems of conventional design require a wide range of imaging bandwidths. Current bandwidths in such systems range from 1 MHz (+/−500 KHz) to approximately 4 KHz. As mentioned above, bandwidth limiting filters are used in such systems in various combinations. In addition to the continuous domain or analog of filters mentioned above, discrete or digital domain filters are also employed following analog-to-digital conversion of the image data. In general, it is desirable to maintain filter parameters such as pass band ripple, transition bandwidth, stop band rejection and so forth relatively constant for all output bandwidths. This provides bandwidth independence for image quality parameters. However, such uniformity is increasingly difficult to realize. For example, in the discrete domain, the transition width provided by a fixed tap length FIR will essentially double with each bandwidth halving in a fixed analog-to-digital rate sample stream. As a result, for low bandwidths a prohibitively large number of taps are required to provide performance equivalent to that at higher bandwidths. Similar performance limitations exist for continuous domain filters.
Due to these limitations in the design of both discrete and continuous domain filters the amount of information filtered out by transition bandwidths and stop band rejection can only be reduced at considerable cost. By consequence, conventional imaging systems typically simply reduce the image field of view by use of a portion of the frequency bandwidth, sacrificing data rejected within the transition region at edges of the filter beyond the low pass filter limit. In many situations, the field of view is thereby reduced by several percent.
It would be useful therefore, to provide a technique for recuperating at least a portion of the image data found in transition regions of the system bandwidth rejected following image data filtering. In particular, there is a need for an improved technique for processing discrete pixel images which economically expands the useful field of view, effectively recuperating such data without the need for a more expensive filter design.
SUMMARY OF THE INVENTION
The invention provides a novel technique for correcting discrete pixel image data designed to respond to these needs. The technique may be applied to a variety of imaging systems wherein the field of view is limited by filters, particularly continuous domain or analog filters. The technique is particularly well suited for use in magnetic resonance imaging systems wherein continuous domain filtering is performed on raw image data prior to conversion of the data to a digital domain for further processing. In accordance with one aspect of the invention, acquired image data is filtered through at least one continuous domain filter to produced filtered data. The filtered data is compiled into discrete frequency value bins including a primary region and a transition region. In general, the transition region is at least partially defined by a non-linear region of the continuous domain filter. The filtered data may be compiled into a large number of such discrete value bins, depending upon the image resolution, field of view, and so forth. A plurality of correction gains are applied to values of the discrete value bins in the transition region to correct the image data.
The filtering of the image data prior to correction may be accomplished in various manners. In one embodiment contemplated by the invention, both continuous and discrete domain filtering is performed prior to the correction of the values in the transition region. Moreover, image data processing prior to the correction step may be performed in the time domain, the frequency domain, or both, such as in magnetic resonance imaging systems wherein image data is acquired in the time domain and subsequently transformed to the frequency domain for image reconstruction. The technique permits expansion of the image field of view in a very economical manner. In certain instances, the field of view is expanded on the order of several percent through the correction technique, while avoiding image artifacts or similar degradation in the imaging process.
REFERENCES:
Wright, G.A., “Magnetic Resonance Imaging”, IEEE Signal Processing Magazine, Jan. 1997, pp. 56-66.
Fletcher Yoder & Van Someren
General Electric Company
Rogers Scott
LandOfFree
Method and apparatus for correction of a discrete pixel image 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 correction of a discrete pixel image, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for correction of a discrete pixel image will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2901741