EP1080377A1

EP1080377A1 - Computer and method for evaluating nuclear magnetic resonance tomography data (turbo-pepsi)

Info

Publication number: EP1080377A1
Application number: EP99926256A
Authority: EP
Inventors: Stefan Posse; Valeri Kiselev
Original assignee: Forschungszentrum Juelich GmbH
Current assignee: Forschungszentrum Juelich GmbH
Priority date: 1998-04-17
Filing date: 1999-04-16
Publication date: 2001-03-07
Also published as: JP2002512099A; DE19817228C1; WO1999054748A1

Abstract

The invention relates to a computer for evaluating nuclear magnetic resonance tomography data, whereby the data contains at least two different relaxation signals of a sample. According to the invention, the computer is designed in such a way that it operates with at least one means of evaluation forming a differential signal from the at least two different relaxation signals, whereby a time characteristic of the differential signal is determined as a function U(t); the computer operates with at least one means of selecting two times, whereby the times can be selected in such a way that the interval ti between both times is shorter than the time characteristic of the differential system and a quotient from the differential signals and one or several noise signals within the interval ti between times t1 and t2 has a greater value in comparison with other time intervals; and the computer has a processing unit enabling approximation of the value of function U(t) within the time interval ti.

Description

description

COMPUTER AND METHOD FOR EVALUATING DATA FROM CORE MAGNETIC RESONANCE TOMOGRAPHY (TURBO-PEPSI)

Technical field

The invention relates to a computer for evaluating

Data from nuclear magnetic resonance tomography, the data containing at least two different relaxation signals from a sample.

The invention further relates to a nuclear magnetic resonance tomograph and a method for evaluating data from nuclear magnetic resonance tomography, relaxation signals from at least two different states of a sample being determined.

Nuclear magnetic resonance (NMR) is used to obtain spectroscopic information about a substance. A combination of nuclear magnetic resonance imaging with techniques of magnetic resonance imaging (Magnetic Resonance Imagmg -MRI) technology gives a spatial picture of the chemical composition of the substance.

State of the art

In medical research in particular there is a need to obtain information about brain activity or, in a broader sense, information about blood flow or changes in deoxyhamoglobin concentrations in animal and human organs. The neuronal activation manifests itself in an increase in blood flow in activated brain areas, whereby there is a decrease in blood Deoxyhamoglobm concentration comes. Deoxyhamoglobin (DOH) is a paramagnetic substance that reduces magnetic field homogeneity and thus accelerates signal relaxation. If the DOH concentration drops due to brain activity that triggers blood flow, the signal relaxation is modulated in the active areas of the brain. The protons of hydrogen in water are primarily excited. Localization of brain activity is made possible by using an investigation using functional NMR methods that measure the NMR signal with a time delay (echo time). This is also known as susceptibility-sensitive measurement. The biological mechanism of action is known in the literature under the name BOLD effect (Blood Oxygen Level Dependent - Effect) and leads to susceptibility-sensitive magnetic resonance measurements with a field strength of a static magnetic field of, for example, 1.5 Tesla up to approx. 10% fluctuations in the Image brightness in activated brain regions. Instead of the endogenous contrast agent DOH, other contrast agents can also occur, which cause a change in the

Cause susceptibility. NMR imaging methods are used to select layers or volumes which, under the appropriate irradiation of high-frequency pulses and the application of magnetic gradient fields, provide a measurement signal which is digitized and stored in a two- or three-dimensional field in the measuring computer.

The desired image information is obtained (reconstructed) from the raw data recorded by a two- or three-dimensional Fouπer transformation.

A reconstructed slice image consists of pixels, a volume data set consists of voxels. A pixel is a two-dimensional picture element, for example a square. The image is composed of the pixels. A voxel is a three-dimensional volume element, for example a cuboid, which - due to measurement technology - does not have any sharp boundaries having. The dimensions of a pixel are of the order of 1 mm ', those of a voxel of 1 mm ³ . The geometries and dimensions can be variable.

Since a strictly two-dimensional level can never be assumed for layer images for experimental reasons, the term voxel (= volume element) is often used here, which takes into account that the image planes have a penetration depth m of the third dimension.

By comparing the measured signal curve in each pixel with the time curve of a model function, a stimulus-specific neuronal activation can be detected and spatially localized. An stimulus can be, for example, a somatosensory, acoustic, visual or olfactory stimulus as well as a mental or motor task. The model function, or the model time series, describes the expected signal change in the magnetic resonance signal as a result of neuronal activation. These can be derived, for example, from the paradigm of the respective experiment using empirical rules. It is important to consider a time delay of the model function compared to the paradigm (slow response of blood flow to neuronal activation).

It is already known how brain activation can be represented by activation images obtained from nuclear tomography data. The activation images can even be calculated and reproduced in real time, which means that a data record can be converted into an image before the next data record is measured. The time interval is typically 1 to 3 seconds.

Such a calculation and reproduction of the

Real time activation images are described in U.S. Patent No. 5,657,758. This method is characterized by from that it enables a high temporal and spatial resolution.

Fast magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS) make it possible to investigate changes in regional blood volumes and blood states depending on an activity in vivo depending on an excitation, see: S. Posse et. al .: Functional Magnetic Resonance Studies of Brain Activation; Seminars in Clmical Neuropsychiatry, Vol. 1, No 1, 1996; Pp. 76-88.

The invention has for its object to further develop the known methods in such a way that the highest possible signal-to-noise ratio is achieved.

Presentation of the invention

According to the invention, this object is achieved by designing a generic computer in such a way that the computer works with at least one evaluation means which forms the difference signal from at least two different relaxation signals, the course of the difference signal over time being determined as a function U (t), that the computer works with at least one means for selecting two times, the times being selectable in such a way that the time interval between them is less than the temporal profile of the difference signal and that within the time interval t _x between times ti and t ₂ em quotient of the difference signal and one or more noise signals has a higher value than other time intervals and that the computer has a computing unit by which the value of the function U (t) is approximated within the time interval t _x . The invention provides to create a computer with which a fast spectroscopic imaging method can be implemented, the changes in the NMR signal relaxation with a

1

Time constant T ₂ = —- at several points in time after one

Suggestion determined.

The spectroscopic imaging method is preferably a spectroscopic echo planar imaging method, in particular a repeated two-dimensional echo imaging method, which consists of a repeated application of a two-dimensional echo planar image coding. Spatial coding takes place in the shortest possible period of time, which is repeated several times during a signal drop and is preferably 20 to 100 ms. By repeating the

Echo-planar coding during a signal drop is shown in the course of the signal drop in the sequence of reconstructed individual images.

An implementation of this particularly advantageous method according to the invention is referred to as TURBO-PEPSI, where PEPSI stands for Proton-Echo-Planar-Spectroscopic-Imag g. An expedient conventional echo planar method, however, is referred to as EPI (Echo Planar Imagmg).

The number of images which are encoded during the signal drop depends on the relaxation time and the encoding time Δt for a single image.

To change the relaxation with the highest possible

Detecting sensitivity was found as a criterion for an optimal choice of the measurement time window as a function of the relaxation time constant, the coding time for a single image and the type of data post-processing. The criterion consists in considering a difference signal between different relaxation states.

The difference signal has a maximum in time, which is close to the mean relaxation time for small relaxation changes.

Preferred evaluation methods, further advantages, special features and expedient further developments of the invention result from the following illustration of preferred exemplary embodiments of the invention on the basis of example calculations, drawings and a table.

Brief description of the drawings

From the drawings shows:

1 shows an experimental difference signal of a functional relaxation time change in a selected picture element as a function of the measuring time after signal excitation,

2 shows a relative, scaled increase in the contrast-noise ratio CNR _N compared to the contrast

Noise ratio CNRi of a single measurement for different evaluation methods depending on the measurements,

Fig. 3 in a first partial image A, a detection of

Brain activation in four steps using a conventional imaging method and in partial image B a detection of brain activation using a method according to the invention,

4 shows a mean correlation coefficient in an activated brain area, in particular in the visual one Cortex, for TURBO-PEPSI depending on the range of the summed echo signals and for the T ₂ ^* image obtained from the TURBO-PEPSI data and a comparison with a conventional echo planar (EPI) method,

Fig. 5 shows a number of activated pixels with a

Correlation coefficients r _mi of at least 0.7 in the visual cortex for TURBO-PEPSI as a function of the range of the summed echo signals and for that

T ₂ ^* image obtained from the TURBO-PEPSI data and a comparison with a conventional echo planar (EPI) method.

The table shows a summary of experimental sample data.

Best way to carry out the invention

The invention provides in particular for detecting a difference signal at different times. These points in time lie within a time interval t_.

In particular, it is a difference signal between a relaxation curve in an excited state and a relaxation curve in a ground state.

An example is a difference signal (vertical axis) between a functional relaxation time change in the human brain in a selected image element in the visual cortex during visual stimulation as a function of the measurement time after signal excitation (horizontal axis) measured by means of fast spectroscopic imaging m Fig. 1. This is a particularly simple case in which the

Difference signal by em difference signal from a relaxation signal during activation and a Relaxation signal is formed during an idle state. However, the term “difference signal” is in no way limited to difference signals, but, like the term “difference function”, includes all cases in which differences between measurement curves are detected or evaluated.

Assuming an exponential drop in the relaxation curves, the difference signal ΔS () shown in FIG. 1 results in:

AS (t) = S ₀ (e- " ^{r {a)} - e ^~ " ^τ ^ ^{b <>} ) ⁽¹⁾

where T ^ (a) and T ₂ ^* (b) are relaxation time constants in an activated state (a) and in a basic state (Baseline State - b) and where S ₀ denotes an output signal intensity.

Assuming a slight change in the relaxation time ΔT ₂ *, the signal difference ΔS (t) is:

(2)

where T- * denotes the relaxation time in the basic state

An essentially bell-shaped curve arises, which has a maximum at t = T ₂ ^" . With a preferred measuring field strength of approximately 1.5 Tesla, t takes a typical value of approximately 70 ms em.

The maximum is:

(3)

A preferred embodiment of the invention provides that it is assumed that the noise effects are so-called white, thermal noise with a mean value close to zero and a standard deviation σ.

The invention proposes various preferred forms of implementation for the evaluation method in order to achieve an increased signal-to-noise ratio compared to a single point measurement. During an Emzel measurement, the contrast-to-noise ratio (CNR) of the formula

corresponds, can with the inventions

Implementation of the evaluation method with an increased contrast-to-noise ratio can be achieved.

A first embodiment of an evaluation method according to the invention provides for the measured effect to be summed for N times and to form an average signal. The average signal gives a good measure of S ₀ T ₂ ^* . Assuming constant measuring intervals Δt for each individual measured value recording and the same noise level in every point, the following applies to the summed signal (t = ι X Δt):

(5)

using the inequalities Δt << T ₂ (6) and N >> 1 (7).

A comparatively small change in T ₂ , as occurs, for example, in the case of blood oxidation (BOLD effect / Blood Oxygen Level Dependent effect), is expressed in the contrast C reproduced below:

10

(8th)

15 where x is defined as follows:

NAt

X

T

20th

(9)

The noise effects in the summed signal according to the formula have the standard deviation:

25th

The contrast-to-noise ratio is as follows

*

1 - (x + l) e -x

₃ ³ ₀ ⁰ CNR N _T = -g2σ Ti 'A ^λ t (ii)

As can be seen, for example, from FIG. 2, the contrast-to-noise ratio has a maximum at x = 3.2. In Fig. 2 that is Contrast-to-noise ratio CNR m as a function of the length of the measurement time after signal excitation T _max , the relaxation rate R2 = 1 / T ₂ ^* and a coding time Δt for various data evaluation methods: summation of the individual measurements (summation), exponentially weighted summation (exponentially weighted summation) , optimally weighted summation (weighted filter, weighted filter) and for curve fitting (fitting).

A maximum contrast-to-noise ratio can be achieved if the measurements are made up to the time

T _max = NΔt = 3.2T ₂ ^* (12).

For an appropriately chosen N, the contrast-to-noise ratio is maximum and is according to the formula:

maximum 0.46.

To further increase the contrast-to-noise behavior, it is expedient to carry out a weighted summation of the signal according to equation 14.

A weighting factor w (t _N ) according to formula 15 is preferably used in formula 14.

w (t _n ) = R2t _n - e ⁿ _d 5) The weighting factor w (t _N ) is an expected relaxation rate in a sample to be examined. This is preferably the average relaxation rate in the sample examined.

The following formula results for the contrast-to-noise ratio:

In this variant of the evaluation method according to the invention, the increase in the signal-to-noise ratio in the multi-point measurement takes a particularly high value of 1.4 em. The measurement time is again preferably 3.2 T _2. With such a weighted summation, it is thus possible to achieve an even better result for the contrast ratio than with a conventional summation.

Another variant of the evaluation method is that an adaptation procedure (fit method) is carried out by adapting the relaxation curve to exponentially falling curves.

The advantage of the evaluation method according to the invention is shown below on the basis of a consideration of the theory of noise effects and on the basis of experiments.

The total signal S _r (t _n ) results as follows:

-R2ß

Here s _Q e ⁿ denotes the pure signal, g _r (t _n ) a white noise and h _r (t _n ) an influence of physiological Disturbance signals of the sample to be examined, which are preferably signals with a low frequency.

The index r assumes values from 1 to NR and means the number of repetitions of the relaxation measurements; the index n takes values from 1 to N and pays the number of echo signals during a relaxation measurement.

In order to extract a change in the relaxation dependency on a brain activation from this measured signal, various approaches, explained using the following formulas, are possible:

A summation over the echo signals results in

while the formula is based on a weighted summation

results, where applies

-R2-t w (t _n ) - R2t _n - e 77

(20)

Another method is a fit method, as shown by the following formula:

As with the general considerations, it can be assumed that the mean value of the white noise is zero or close to zero. The contrast-to-noise ratio (CNR) results from ΔS divided by the total noise. The difference value is then determined for at least two measurements.

According to another preferred embodiment of the invention, a correlation analysis is carried out for each individual echo signal over a plurality of relaxation measurements taking place successively in time. The correlation analysis is carried out in a known manner, an implementation according to the article by Peter A. Vandettmi et al. in: Magnetic Reonance in Medecme, vol. 30, pp. 161-173, 1993, to which full reference is made, is particularly expedient.

The expected value for the correlation coefficient (c.c.) is

σ 2

(c.c.) = c.c. 1 - (22) where

0

2AS 2

De 2 / correlation coefficient f l z ient c. c. has a standard deviation

on. This is followed by a combination of the correlation coefficients, for example by averaging.

An experimental check of the evaluation method according to the invention was carried out on the basis of nuclear magnetic resonance imaging examinations of the brains of test subjects. A light source, in particular a matrix of luminescent diodes (light emitting diode LED), was positioned in the immediate vicinity of the test subjects' faces and stimulated to produce signal flashes. The excitation frequency is 8 Hz. The action of the signal flashes takes place via a time interval of several seconds, for example 5 seconds, synchronized with a carrier signal of a scanner, which is followed by an approximately equally long idle interval. The scanner is a Vision 1.5 Tesla total body scanner from Siemens Medical Systems, Erlangen, in standard equipment with a magnetic field gradient of 25 mT / m. Such a scanner is able to switch gradient fields within approximately 300 μs.

TURBO-PEPSI (Proton-Echo-Planar-Spetroscopic-Imagmg) was used as the spectroscopic imaging method.

Data is adjusted according to the exponential function:

*

(~ TE / T ₂ *) s 2 0 ^C (25),

using a non-linear least-square fit.

Of voxels in which the signal intensity at the first echo is 10% of the maximum measured in the entire image

Signal amplitude exceeded and where the

Correlation coefficient between the measured date and the fitted data exceeded 0.95, were parametric Images of T ₂ , the output signal amplitude S ₀ and χ ^{2 are} formed.

These parameters were set to 0 in the other voxels. With the use of these criteria, with the exception of the ventricles, excellent fit of the fitted data to the experimental results was achieved in all brain regions. In most voxels, the correlation coefficient exceeded 0.99.

Alternatively, the echoes of each relaxation measurement are averaged and then a correlation analysis is carried out both for the parametric images and for the averaged biloers.

The experiments showed extensive activation areas of the primary visual cortex (V ₁ ) as well as in adjacent regions (V ₂ ) of the visual cortex.

The mean correlation coefficient in this region (Vi,

V ₂ ) and the number of averaged echo signals are shown in FIG. Δ m in comparison to a conventional EPI method.

In Fig. 5, the number of activated pixels is one

Correlation coefficients of at least 0.7 m as a function of the number of averaged echo signals are shown.

A summary of the experimental results from Figures 4 and 5 can be found in the table on page 19.

The invention provides a computer for evaluating data from nuclear magnetic resonance tomography, a nuclear magnetic resonance tomograph equipped with the computer, and a method for evaluating data from nuclear magnetic resonance tomography. The invention has a number of advantages. This includes optimizing the measurement sensitivity for a quantitative measurement of the relaxation time and the qualitative change in relaxation time. This makes it possible to use imaging with the highest possible bandwidth (shortest coding time) for the least possible spatial distortion and to achieve maximum measurement sensitivity by measuring an optimal number of codings after signal excitation.

The evaluation method can be used in real-time measurements and analyze the relaxation changes there.

Furthermore, the evaluation methods according to the invention are particularly versatile. It has proven to be expedient to use a summation or, what is even more advantageous, a weighted summation, which can be done at a higher speed and without loss of sensitivity compared to curve fitting. A

Summation or a weighted summation have the advantage that they represent a particularly robust evaluation method.

It is also possible with the help of the invention in all

To achieve an optimal adaptation of the measuring sensitivity, especially with measuring field strengths of 0.1 Tesla to 15 Tesla, for example by selecting the number of echo signals as a function of the relaxation time, the number preferably being chosen according to formula 12.

All subjects showed strong activation in the primary visual cortex (Vi) and adjacent areas. The observed changes in the

TURBO-PEPSI measured functional signal ranged from 3 to 20% depending on the echo time, the location and the respective subject. The excitation has a maximum near TE = T ₂ ^* . When EPI and TURBO-PEPSI images were compared with TE = 72.5 ms, very similar activation images were determined.

When using a correlation limit of 0.4, it was also possible to detect smaller signal changes with echo times of, for example, 12.5 ms to 228 ms. Averaging the correlation images reduces the intensity of noise effects compared to EPI. The spatial extent of the activation zone and the increased correlation coefficients in the visual cortex increase with the number of echoes added up, as can be seen from FIGS. Experiments with longer excitation times (7 to 12 seconds) become larger

Correlation coefficients obtained than in measurements with shorter excitation times (for example 3 seconds). In the first, images with similar correlation coefficients and activation zones as in the T ₂ ^* images result - see FIGS. 4 and 5. It can be seen that a particularly high sensitivity is achieved by summing the first, preferably the first 6 to 10, in particular the first 8 echo signals corresponding to the plateau of the CNR curve m Fig. 2 can be achieved.

The sensitivity gain is particularly advantageous for real-time measurements, because a change in relaxation can be effectively determined even with a few measured values. In summary, it can be said that by multi-echo detection of the difference signal an optimal sensitivity is achieved at any magnetic field strength.

In addition, the invention is both in echo planar imaging (echo planar imaging EPI) and in phase-coded

Imaging methods as well as spectroscopic imaging methods can be used. An increase in the sensitivity to measurement by about 40% to 140% according to the invention results in a significant increase in the sensitivity to measurement in a large number of application areas, for example in the cognitive or visual activation of the human brain.

The examples shown explain the computer and the evaluation method using NMR measurements on the human brain. Of course, both the computer and the magnetic resonance tomograph as well as the evaluation method can also be used to examine other samples of living or non-living material.

Range of increase in Z-p-value increase in p-value averaged transformed counteractivated gegen¬

Echo in the visu over areas (r _mιn = over

Signals cortex of a 0.7) [%] (SD) of a

V1 / V2 [%] single versus single versus single echo measurement

Single echo measurements

Measurement (SD)

4-5 9 (7) 0.018 -3 (13) 0.585

3-6 38 (14) 0,000 176 (156) 0.025

1-8 51 (26) 0.002 286 (243) 0.021

1-10 56 (25) 0.003 295 (223) 0.023

1-12 60 (24) 0.002 304 (240) 0.027

Claims

claims

1. Computer for evaluating data from nuclear magnetic resonance tomography, the data containing at least two different relaxation signals of a sample, characterized in that the computer works with at least one evaluation means that forms a difference signal from at least two different relaxation signals, wherein the time course of the difference signal is determined as a difference function U (t), - that the computer works with at least one means for selecting two times, the times being selectable in such a way that the time interval t _∑ between them is less than the time profile of the difference signal and that within the time interval ti em between the times tη and t-, the quotient of the difference signal and one or more noise signals has a higher value than other time intervals and that the computer has a computing unit by means of which the value of the difference function ion U (t) within the time interval t _. is approximated.

2. Computer according to claim 1, d a d u r c h g e k e n n z e i c h n e t that it contains em means for evaluating the difference function as contrast C.

3. Computer according to one of claims 1 or 2, characterized in that the computing unit relaxation signals according to the formula Sr = (t _n ) calculated, where S _r (t _n ) em signal to one

Designated time t _n .

4. Computer according to one of claims 1 or 2, d a d u r c h g e k e n n z e i c h n e t that the computing unit relaxation signals according to the formula

_ N

S _r = ∑S _r (t _n ) - w (tA) calculated, where n- \

- / ?? • / w (t _n ) = R2t _n -e ⁿ , where S _r (t _n ) em signal to one

Designates time t _n and where w (t _N ) is a weighting factor into which an expected relaxation rate of a sample to be examined is included.

5. Computer according to one of claims 1 or 2, d a d u r c h g e k e n n z e i c h n e t that the computing unit relaxation signals according to the

formula

S _r - calculated, where

-R2-t s _Q denotes a pure signal and S _r (t _n ) em signal at a time t _n

6. Computer according to one of claims 1 to 5, d a d u r c h g e k e n n z e i c h n e t that the computing unit the difference function according to the

formula

N

U _r - ∑ Ur (t _n ) -w (t _n ) calculated, where U _r (t) em n = \

Differential signal designated and where applies - / ?? • / w (t _n ) = R2t _n ^■ e ⁿ and where w (t _N ) e is a weighting factor, which receives an expected relaxation rate of a sample to be examined.

7. Computer according to one of claims 1 to 5, d a d u r c h g e k e n n z e i c h n e t that the computing unit the difference function according to the formula

_ N Ur = ∑U _r (t _n ) calculated, where U ₁ (t _n ) em n = \

Differential signal called.

8. Nuclear magnetic resonance scanner, so that it contains at least one computer according to one of claims 1 to 7.

9. A method for evaluating data from nuclear magnetic resonance tomography, wherein relaxation signals from at least two different states of a sample are determined, characterized in that em difference signal is formed from different relaxation signals that em time course of the difference signal is determined as a function U (t) and that two

Times are selected, the selection of the times ti and t ₂ taking place in such a way that in a time interval between them t _± em ratio between the difference signal and one or more noise signals is greater than in other time intervals and that within the time interval t _± the value the difference function U (t) is determined.

10. The method according to claim 9, characterized in that the method is carried out several times, the values for t_ and t _{2 being} varied on the basis of one or more previous process runs in such a way that the ratio between the difference signal and the noise signal is as large as possible.