Electricity: measuring and testing – Particle precession resonance – Using a nuclear resonance spectrometer system
Reexamination Certificate
2000-07-18
2002-05-14
Lefkowitz, Edward (Department: 2862)
Electricity: measuring and testing
Particle precession resonance
Using a nuclear resonance spectrometer system
C324S307000, C324S300000
Reexamination Certificate
active
06388443
ABSTRACT:
BACKGROUND OF THE INVENTION
The invention relates to a computer for evaluating signals from nuclear magnetic resonance tomography and a nuclear resonance tomograph equipped with such a computer.
In medical research, it is desirable to obtain information with respect to the brain activity or, in a wider sense, information concerning the blood flow or the deoxyhemoglobin concentration changes in animal and human organs. The neural activation results in an increase of the blood flow in activated brain areas, which causes a decrease in the blood deoxyhemoglobin concentration. Deoxyhemoglobin (DOH) is a paramagnetic compound which accelerates the nuclear spin-dephasing. If the DOH concentration is lowered because of a brain activity which causes a blood flow, the dampening effect on the NMR radiation in the active areas of the brain is reduced, so that NMR signals (nuclear magnetic resonance signals) of the water contained in the brain have a higher intensity. Mainly the protons of the hydrogen in the water become excited. DOH consequently has a dampening effect on the NMR radiation. It is reduced as a result of brain activity and consequently, permits a localization of the brain activity if examinations utilizing NMR methods are performed. This action mechanism is known in the field under the name BOLD effect (Blood Oxygen Level Dependent-effect) and results in susceptibility-sensitive magnetic resonance measurements at a field strength of a static magnetic field of for example 1.5 Tesla in up to about 10% changes of the image brightness in the activated brain regions. The endogenic contrast compound DOH may be replaced by other contrast compounds, which cause a change in the susceptibility. With NMR imaging methods, layers or volumes are selected which, with suitable input of high-frequency radiation pulses and the application of magnetic gradient fields, provide a measuring signal, which is digitized and stored in the measuring computer in a two- or three-dimensional field.
Normally, FFT is used, but the use of other transformations might also be used. From the recorded raw data, the desired image information is reconstructed by way of a two- or three-dimensional Fourier transformation.
A reconstructed slice image consists of pixels, a volume data set consists of voxels. A pixel is a two-dimensional image element, for example a square, of which the image is composed. A voxel is a three-dimensional volume element, for example a block which, because of measuring techniques, has no sharp limit. The dimensions of a pixel are in the range of 1 mm
2
, those of a voxel are in the range of 1 mm
3
. The geometries and dimensions are variable.
Although the images of the measured slices are only two-dimensional, the term “voxel” is also used for them By a comparison of the measured signals in each pixel with the time-dependent plot model function, a stimulus-specific neural activation can be detected and localized. A stimulus may be for example a somato-sensoric, an acoustic, a visual or an olefactoric stimulus as well as a mental or motor task. The model function or, respectively, model time series describes the expected signal change of the magnetic resonance signal as a result of neural activation. Those may be derived or example by empirical rules from the paradigm of the respective experiments. It is essential to take into consideration a time delay of the model function with respect to the paradigm (slow reaction of the blood flow to neural activation).
It is already known how brain activation can be represented by images, which were derived from nuclear spin tomographic data. The computation and the display of the activation images are even possible on a real-time basis that is an image can be created from a data set before the next set of data is obtained. The time gap is typically 1 to 3 seconds.
The most widely used method of detecting neural activity is the correlation analysis. It is realized by computation of the cross-correlation coefficients between a reference vector, that is a model time series, and the time sequences of the voxels taken into consideration. The real time capability of a correlation algorithm requires that the time needed for the renewed computation of the correlation coefficients is constant for each new set of magnetic resonance data recorded.
A real time correlation analysis is known from Cox, R. W., Jesmanowicz, A., J. S. Magn. Reson, Med., 33 230, 1955, which supports the suppression of low-frequency noise by means of a detrending procedure. In a detrending procedure, one tries to reduce the effect of non-stimulus induced signal changes into the time sequences measured. This means mathematically that the measurement vector, that is, the vector which includes the measured time sequence of a voxel is split into the sum of two orthogonal vectors. The part, which is described by a linear combination of detrending vectors, is discarded. The detrending vectors, which in a mathematical sense form a basis, include the time series by whose weighted sum the complete low frequency noise part of the measured time series is described (“complete” within the limits of the given detrending vectors). By applying detrending to the measuring and reference vectors entering with the correlation computations, the effect of non-stimulus induced signal changes on the correlation images can be reduced.
With the application of detrending, the correlation coefficient &rgr; is calculated according to the formula (1).
ρ
=
x
→
S
r
→
S
&LeftBracketingBar;
x
→
S
&RightBracketingBar;
&LeftBracketingBar;
r
→
S
&RightBracketingBar;
(
1
)
Wherein:
&rgr;=correlation coefficient
{right arrow over (x)}
s
=measurement vector after detrending
{right arrow over (r)}
s
=reference vector after detrending
Furthermore, Cox, R. W. presented in Comput. Biomed. Res. 29, 162, 1996, the program AFNI, which is based on the algorithm of the first-mentioned publication. The known algorithm is based on elements of a matrix which is obtained by Cholesky analysis of another matrix which is composed of the scalar products of the measurement and reference vectors as well as the detrending vectors.
The neural activation measured with a nuclear spin tomographic examination follows the stimulation paradigm, that is the protocol, which represents the timing of the stimuli. Usually an on-off period (that is, a stimulus “on” phase followed by a stimulus “off” phase) is repeated several times.
If a correlation analysis takes all measuring time points into consideration, the sensitivity for changes of the signal-response form decreases with each activation cycle. However, the prior art methods do not provide for a uniform sensitivity for the change of the signal response form.
It is therefore the principal object of the present invention to provide a computer and a nuclear magnetic resonance tomograph, which deliver improved analysis results. It is a particular object to provide for a uniform sensitivity for the changes of the signal response form. Also, a computer and a nuclear magnetic resonance tomograph are to be provided which permit the implementation of such a solution. For this purpose, a new detrending procedure for the reduction of undesired low-frequency components in the measured signals is to be developed, whereby the implementation is made possible.
SUMMARY OF THE INVENTION
In a computer for evaluating signals from nuclear magnetic resonance tomography and a tomograph using such a computer, the neuronal activity measured during a nuclear magnetic resonance tomographic examination follows a protocol, which describes the timing of the stimuli. During a correlation analysis, the sensitivity of known means to changes in the signal-response form normally decreases with each activation cycle. According to the invention, better evaluation results can be achieved in the correlation analysis by using a constant number of n data values in a correlation analysis employing a sliding-window technique. As the data measurement progresses,
Gembris Daniel
Posse Stefan
Taylor John
Bach K. J.
Forschungszentrum Jülich GmbH
Lefkowitz Edward
Shrivastav Brij B.
LandOfFree
Computer for evaluating signals from nuclear magnetic... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Computer for evaluating signals from nuclear magnetic..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Computer for evaluating signals from nuclear magnetic... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2896861