JPH08131414A - Sampling method for living body function invigoration information in magnetic resonance imaging - Google Patents

Abstract

PURPOSE: To sample the invigoration information of a cranial nerve, etc., with high precision even when no model is established by measuring a magnetic resonance signal in time series fashion, finding the time differentiation of the time series data of the magnetic resonance signal at every pixel and sampling an invigoration part based on the time differentiation. CONSTITUTION: An MR signal is measured by laying down a patient 4 on a bed 5 set in a uniform magnetostatic field made of magnet 1, transmitting a high-frequency magnetic field by an RF coil 3, and receiving the high-frequency magnetic field discharged from the patient 4 with the RF coil 3 after such transmission is stopped. At this time, position information is superimposed on the high-frequency magnetic field received by the RF coil 3 by applying a spatially inclined magnetic field by a coil 2. The high-frequency magnetic field measured in such way goes to the MR signal. In such a case, the magnetic resonance signal is found in time series fashion, and the time differentiation of the time series data of the magnetic resonance signal is found at every pixel, and the invigoration part can be sampled based on the time differentiation.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for extracting biofunction activation information of activation sites such as cranial nerves and activation start times from the results of biofunction measurement by magnetic resonance imaging (MRI).

[0002]

2. Description of the Related Art Tomographic imaging (MRI) utilizing nuclear magnetic resonance is widely used in the field of medical diagnosis. M
Image information provided by RI was mostly anatomical, but in 1992, a method of measuring brain function (Proc. Natl.
Acad. Sci. USA, 1992, vol.89, pp.5951-5955), it has been attracting attention as one of the brain function information. In brain function measurement by MRI, changes in the components of venous blood and blood flow that accompany the activity of cranial nerves are observed as changes in magnetic permeability. The blood vessels of interest here penetrate deep into the brain and are minute, and it is thought that changes in venous blood are strongly related to the activity of nerve cells in the vicinity. Therefore, in order to measure the brain function by MRI, the data (function data) when activating the cranial nerve and the venous blood reference are used as time-series data (MR data) of two-dimensional or three-dimensional MR signals. It is necessary to measure the data (reference data) when the cranial nerve that becomes The difference between the functional data and the reference data is venous blood change, that is, brain functional information.

It takes a lot of time for the measurer to find out the change in venous blood from the MR data and extract the cranial nerve activation site. Therefore, a method of automatically and objectively extracting a changed portion of venous blood has been proposed. For example, there is a method of conducting a statistical test between functional data and reference data to search for a site having a significant difference, and a method of correlating MR data with the period of stimulation application and rest (Proc.
eedings of the Society of Magnetic Resonance in Me
dicine 1993, p.449 & p.1376). In the extraction method using correlation, the MR data obtained during the stimulation application period corresponds to the functional data, and the MR data obtained during the stimulation rest period corresponds to the reference data.

[0004]

The extraction method described above assumes a model in which there is no change in venous blood in the reference data and the change occurs in the functional data.
However, such a model does not always hold in an actual activation site, and if the model does not hold, an erroneous result may be derived. An object of the present invention is to provide a method capable of accurately extracting a venous blood change site and a change start time, that is, activation information such as a cranial nerve, even when the model does not hold.

[0005]

In the present invention, the a priori assumption is used by calculating the time derivative of the MR data for each pixel and extracting the peaks exceeding the threshold value from the time series data of the time derivative value. Without doing so, the presence or absence of activation of the cranial nerve and the like is determined, and the site and start time are extracted to achieve the above object. Also, as the activation start time, the time of the earliest time change of the MR data is used.

The time derivative of MR data can be obtained from the difference between the signal strength at a certain time and the signal strength at a time preceding or succeeding the time. This difference is
A large peak is shown at the time when a change appears in the MR data.
Therefore, when this peak is extracted from the time series data of the time differential value, the time and magnitude of the time change of the MR data can be known. However, since there is noise in the actual data, a threshold is set and only peaks with a size exceeding this threshold are extracted. MR data are preferably obtained from magnetic resonance signals of hydrogen nuclei.

[0007]

[Function] MR data containing functional information includes temporal changes in signal intensity due to changes in blood flow of venous blood or changes in blood components associated with activities such as cranial nerves caused by stimulation. Exists. Therefore, by monitoring the time change of the MR data for each pixel of the image composed of the MR data, it is possible to extract the activation site such as the cranial nerve.

The method of the present invention does not need to assume a model of cranial nerve activity accompanying stimulus presentation or task execution, and therefore can be used when the model is not valid or the model cannot be confirmed to be valid. In addition, since the differential processing has a low-frequency cutoff characteristic, it can be used without preprocessing for data including a baseline change. Furthermore, it is possible to easily examine the latency from the start of the stimulus or task until the change in the MR signal appears, and the duration of the signal change.

[0009]

EXAMPLES The method of the present invention is shown in FIGS. 1 (1) to (6).
It is executed according to the procedure. An example will be described below according to the procedure. (1) Measurement of MR data FIG. 2 is a configuration diagram showing an example of a brain function measuring device using MRI. In the figure, 1 is a magnet that creates a static magnetic field, 2
Is a gradient magnetic field generating coil that creates a magnetic field with a spatial gradient, 3
Is an RF coil for transmitting and receiving a high frequency magnetic field, 4 is a subject, 5
Is a bed that supports the subject, and 6 is a workstation for data collection and analysis.

In order to measure the MR signal, the subject 4 is laid down on the bed 5 in a uniform static magnetic field created by the magnet 1 and the RF is applied.
The high frequency magnetic field is transmitted by the coil 3 for a certain period of time, and after the transmission is stopped, the high frequency magnetic field emitted by the subject 4 is received by the RF coil 3. At this time, a spatially inclined magnetic field is applied to the coil 2
Position information is superimposed on the high frequency magnetic field received by the RF coil 3. The high-frequency magnetic field measured in this way becomes an MR signal. Depending on the pulse sequence used for measurement, in order to obtain an MR signal as one piece of image data, it is necessary to repeat the process of transmitting a high frequency magnetic field, applying a gradient magnetic field, and receiving a high frequency magnetic field while changing the magnitude of the gradient magnetic field. is there. M which is time series data of MR signal
In order to obtain R data, the process of collecting MR signals is repeated as many times as necessary to obtain brain function information.

In the brain function measurement, for example, as shown in FIG. 3, MR data is collected during a rest period in which no stimulus is presented to the subject or no task is performed, and a stimulus period in which a stimulus is presented or a task is performed. In Figure 3, M
Two-dimensional MR signals are acquired 7 times at equal time intervals to obtain R data.

(2) Limitation of analysis region Next, the region of the MR data from which activation information is extracted is limited to the portion where the head exists. In that case, the difference in MR signal intensity between the part where the head is present and the background part is used, and a threshold is set for the intensity to distinguish the brain from the background. In addition, it is possible to distinguish from the brain by setting a threshold value for a portion of the subject's head other than the brain, such as the scalp and skull. Note that this process may be omitted depending on the data. Furthermore, the MR data is decomposed for each pixel. Time-series M obtained from pixels containing brain function information
An example of R data is simplified and shown in FIG. In the MR data of FIG. 3, there is a first change at the start of stimulation and 2 at the stop of stimulation.
There is a second change.

(3) Pre-processing The time-series data generated in the processing of (2) is subjected to processing for removing high frequency components for each pixel. This processing is realized by a digital filter. Note that this process may be omitted depending on the data. (4) Differentiation processing For each pixel, time differentiation processing of the time-series data for which the preprocessing (3) has been completed is performed. This process consists of data at a certain time k (kth data) and data (k) before time α.
-Αth data) or data (k + αth data) after time α after time k. Also,
As shown in the following formula, data of times before and after that may be used.

D (k) = a (0) .S (kn) + a (1) .S (k-n + 1) + ... + a (n) .S (k) + a (n + 1). S (k + 1) + ・ ・ ・ + a (2n-1) ・ S (k + n-1) + a (2n) ・ S (k + n) (1) where D (k): at time k Differential value a (m), (m = 0 to 2n): Weighting coefficient S (k + l), (l = -n to n): Time (k + 1) data n: Positive integer, (2n + 1) Is the number of data used for the differential operation.

For example, when the differential of time k is calculated as the difference between the data at time k and the data at time (k-1) immediately before that, the parameter of equation (1) is n =
1, a (0) =-1, a (1) = 1, a (2) = 0, and when calculating as the difference between the data at time k and the data at time (k + 1), which is one time after that. , N = 1, a (0) =
0, a (1) = 1, a (2) = − 1. FIG. 4 schematically shows the differential processing process using the MR data shown in FIG.

(5) Peak Search From the time series data after the differential processing (4),
As shown in, the data that exceeds the threshold set by the measurer is searched, and the corresponding data is recorded as a peak (activation peak) due to a change in the MR signal accompanying activation of the cranial nerve. (6) Display of results Whether or not there is a signal change can be found for each pixel by peak search. If there is a signal change, information about the time when the signal change occurred and the direction in which the change occurred can be obtained.
Furthermore, if there are two signal changes with different directions in one MR data, the MR
It also gives information about the time the data was changing.
The time of the first activation peak can be used as the start time of the change of the MR signal.

Information on the magnitude of the signal change (signal change rate) and the time change of the signal change rate can be obtained from the unprocessed MR data and the peak time of the time derivative thereof. Note that these pieces of information can also be obtained by integrating the time derivative of MR data. For the measurer,
These pieces of information obtained for each pixel are presented on the workstation 6 as a difference in color or pattern on the image. FIG. 6 shows an example thereof. In the figure, the pixels to which the pattern is attached indicate that there is a signal change in the time-series data and that they are cranial nerve activation sites.

When the activation start time is extracted using the time derivative, the time derivative of the time derivative of the MR data is performed in the derivative process (4), and the obtained time derivative value is searched for a peak (5). The result is displayed (6) using the time of the first activation peak as the activation start time. If the activation site has already been extracted, this operation can be limited to the spatial coordinates.

Although the present embodiment deals with two-dimensional MR data, it is possible to handle three-dimensional data with substantially the same procedure. The differences are that MR data is collected in three dimensions, pixel-by-pixel processing becomes voxel-based processing, and the results are displayed three-dimensionally or multiple tomographic images as shown in FIG. 6 are displayed. Is. It should be noted that here, the extraction of the activation information of the cranial nerve is described as an example,
The present invention can also be applied as a method for extracting nerve activation information of peripheral nerves and organs other than the brain and changes in venous blood.

[0020]

EFFECTS OF THE INVENTION In the present invention, unlike the method using the correlation or significant difference test which is a known extraction method, it is not necessary to presume a model of cranial nerve activity accompanying stimulus presentation or task execution. It can also be used when it is not valid or the model cannot be validated. In addition, since the differential processing has a low-frequency cutoff characteristic, it can be used without preprocessing for data including a baseline change. Furthermore, since the signal change time is easily extracted, the latency from the start of the stimulus or task to the change in the MR signal and the duration of the signal change can be easily examined.

[Brief description of drawings]

FIG. 1 is a flow chart showing an automatic extraction procedure of activated areas of cranial nerves.

FIG. 2 is a configuration diagram showing an example of a biological function measuring apparatus using MRI.

FIG. 3 is a schematic diagram of measurement results of MR data in brain function measurement.

FIG. 4 is a schematic diagram of time differential processing of MR data.

FIG. 5 is a schematic diagram of a time differential peak search.

FIG. 6 is a diagram showing a display example of activated areas of cranial nerves.

[Explanation of symbols]

1 ... Magnet that creates a static magnetic field, 2 ... Gradient magnetic field generating coil, 3 ...
RF coil, 4 ... Subject, 5 ... Bed, 6 ... Workstation for data collection / analysis

Claims (6)

[Claims] 1. A method of extracting biofunction activation information from data obtained by magnetic resonance imaging, wherein magnetic resonance signals are measured in time series, and time derivative of time series data of magnetic resonance signals is obtained for each pixel. A method for extracting biofunction activation information in magnetic resonance imaging, which comprises extracting activation sites based on the time derivative. 2. The method for extracting biofunction activation information according to claim 1, wherein the biofunction is a function of a cranial nerve. 3. The size of the time differential value of the time series data of the magnetic resonance signal obtained for each pixel is compared with a preset threshold value, and the presence or absence of activation is determined for each pixel by the presence or absence of the time differential value exceeding the threshold value. The method for extracting biofunction activation information according to claim 1 or 2, characterized in that 4. A weighted addition including one or more positive and negative weighting factors for a magnetic resonance signal obtained at a predetermined time and one or a plurality of magnetic resonance signals obtained before and after the time. The method for extracting biological function activation information according to claim 1, 2 or 3, wherein the first-order time differential value of the magnetic resonance signal at a predetermined time is calculated by performing. 5. The method for extracting biofunction activation information according to claim 3, wherein the earliest time of the biofunction activation times determined in the determination of activation is extracted as the activation start time. 6. A time derivative of a time derivative of time series data of a magnetic resonance signal is obtained for each pixel, the value is compared with a preset threshold value, and the earliest time exceeding the threshold value is extracted as an activation start time. 6. The method for extracting biofunction activation information according to claim 5.