Image analysis – Image enhancement or restoration
Reexamination Certificate
1998-11-17
2002-01-22
Lee, Thomas D. (Department: 2624)
Image analysis
Image enhancement or restoration
C382S255000
Reexamination Certificate
active
06341179
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to image processing and in particular to image processing techniques whereby image artefacts produced during image acquisition are corrected for in the generation of an output image.
2. Discussion of Prior Art
Magnetic Resonance Imaging (MRI) is a widely used technique for medical diagnostic imaging. In a conventional MRI scanner, a patient is placed in an intense static magnetic field which results in the alignment of the magnetic moments of nuclei with non-zero spin quantum numbers either parallel or anti-parallel to the field direction. Boltzmann distribution of moments between the two orientations results in a net magnetisation along the field direction. This magnetisation may be manipulated by applying a radiofrequency (RF) magnetic field at a frequency determined by the nuclear species under study and the strength of the applied field. In almost all cases, the species studied is the nucleus of the hydrogen atom, present in the body mainly in water molecules, and RF pulses are applied at the resonant frequency of these water protons.
The energy absorbed by nuclei from the RF field is subsequently re-emitted and may be detected as an oscillating electrical voltage, or free induction decay (FID) signal in an appropriately tuned antenna. More commonly, a further RF pulse, or magnetic field gradient, is used to postpone signal acquisition and generate a spin-echo or gradient-echo signal.
Spatial information is encoded into the echo signal by virtue of additional linearly varying magnetic fields, known as magnetic field gradients, applied during or prior to echo acquisition. The principle of spatial encoding is that in the presence of the field gradient, the net field experienced by a given nuclear moment, and hence its resonant frequency, is a function of position within the scanner. When a gradient is applied during the echo acquisition, the received signal contains a range of frequency components representing nuclei at different locations along the gradient direction. Fourier transformation of this signal yields a one-dimensional projection through the patient. This technique is known as frequency encoding. Two-dimensional encoding requires use of an additional gradient applied perpendicular to the frequency encoding axis, known as the phase encoding gradient. This gradient is applied for a short time prior to data acquisition. The acquisition process is repeated perhaps 256 or 512 times using phase encoding gradients of different strengths. Simultaneous frequency and phase encoding yields a two-dimensional data set which when subjected to two-dimensional Fourier transformation provides the required image. This array of data exists in what is known as k-space and is the Fourier transform of the image space. The effect of the phase encoding gradient is to move the start of the data acquisition to a particular location along one axis in k-space (dependent on the gradient strength), whilst frequency encoding represents a sweep through k-space parallel to the other axis. Each of these sweeps is known as a “shot” or “view”.
Spatial localisation in the third dimension may be achieved using an additional phase encoding gradient, or more commonly by using a gradient and narrowband RF pulse to restrict the initial perturbation of nuclear moments to a single tomographic slice. This principle can readily be extended to multislice MRI.
In conventional MRI, a single phase-encoding view is acquired after each RF excitation. However, faster imaging sequences now exist in which further RF pulses and phase encoding gradients are used to acquire a train of differently encoded echoes after each excitation. These echoes traverse several lines of k-space and reduce scanning time by a factor equal to the echo train length. In the extreme case, single shot echo planar imaging (EPI) techniques cover the whole of two dimensional k-space in a single acquisition lasting less than 100 ms, although spatial resolution and image quality may be significantly compromised.
Patient movement during the acquisition of MRI images results in degradation of the images that can obscure the clinically relevant information. Each readout period takes a few milliseconds (ms), whereas the time interval between readouts might be between 500 and 2000 ms. The majority of blurring and ghosting artefacts caused by patient motion are due to motion between lines in k-space, rather than motion during a single readout.
Movement leads to phase errors between lines of k-space, which in the resulting image appear as blurring and ghosting along the phase encode direction. These phase errors can result from translations and rotations. Translations of the patient in the readout direction result in a frequency dependent phase shift in each line of k-space. Rotations in the spatial domain are also rotations in k-space, and result in a phase shift that is a more complicated function of position.
A particular type of MRI image investigation, known as diffusion weighted imaging, takes place in the presence of an additional and separate gradient. The integral over time of the diffusion weighted gradient is greater than the integral for the phase encode or readout gradients. The purpose of this additional gradient is to make the images sensitive to molecular motion of the order of 10 &mgr;m. A side effect is that the images are also sensitive to bulk motion on the same scale. Anderson et al. in Magn. Reson. Med. Volume 32, 1994, pages 379-387 have shown that for small rigid body movements, the resulting artefacts can be modelled using zero and first order phase correction terms.
Considerable work has been done by MR researchers to model patient motion, and to attempt to correct for it. The impact of different types of motion on the resulting images is well understood, but clinically usable retrospective motion correction techniques are not yet available. Existing algorithms tend to correct only for one dimensional motion, or they require exotic image acquisition strategies that are not generally applicable.
There are broadly two classes of movement correction algorithm used in MRI; with and without “navigator echoes”. Motion correction using additional echoes referred to as “navigator echoes” involves the acquisition of additional echoes that are not phase encoded between each phase encoded echo. All navigator echoes are projections through the object. It is therefore possible to measure motion between the navigator echoes, and consequently infer the motion between the corresponding phase encoded echoes. The navigator echoes are most commonly used to measure motion in one or more translational directions. Published papers describing the use of navigator echo techniques include that of Ehman et al. in Radiology, Volume 173, 1989, pages 255 to 263. Recently several authors have proposed obtaining rotational information either from these straightforward navigator echoes, for example Anderson et al. in Magn. Reson. Med. Volume 32, 1994, pages 379-387, or from circular navigator echoes for example Fu et al. in Proc. Soc. Magn. Reson., 1994, page 355.
There have been attempts to measure movement directly from the phase encoded data. Felmlee et al., as reported in Radiology, Volume 179, 1991, pages 139 to 142 tried to measure translations directly from a hybrid space comprising the Fourier transform of the readout vs. phase encode, but found that it worked for phantoms with high spatial frequency edges, but on human subjects only if high contrast markers were used. A possible solution to this is to acquire spiral readouts, all of which sample a range of spatial frequencies as described by Khadem et al. in Proc. Soc. Magn. Reson., 1994, page 346, but this is impractical on the majority of MRI hardware. An alternative strategy described by Wood et al. in J. Magn. Reson. Imag., Volume 5, 1995, pages 57 to 64, is to locate discontinuities in k-space that correspond to sudden movements of the patient, to split the regions of k-space between these discontinuities into s
Atkinson David
Hill Derek L G
Stoyle Peter N R
Brinich Stephen
Lee Thomas D.
Nixon & Vanderhye P.C.
The Secretary of State for Defence in Her Brittanic Majesty&apos
LandOfFree
Method and apparatus for imaging artefact reduction 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 imaging artefact reduction, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for imaging artefact reduction will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2822483