JP2009273539A - Method for extracting state variation from imaging data, system for visualizing state variation, and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an imaging data processing method for intuitively understanding how state variation points change on imaging data of an object to be measured which has been measured, a computer program, and its operating method. <P>SOLUTION: When data obtained by measuring time-variable imaging data in still images of signals emitted by the object to be measured in a stationary state at a plurality of times is acquired, the object to be measured is partitioned into a plurality of regions on the imaging data for data of each measurement, and a model function of a stationary state is generated for each section. Then, time-series residual data for each section is calculated by using the generated model function of the stationary state, and a statistical test of is performed for the residual data. The imaging data processing method for mapping the calculated examination values with respect to the prescribed section at prescribed time when calculating the examination value equal to or higher than a prescribed level of significance, the computer program, and its operating method are provided. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a method for processing imaging data measured in time series at observation points arranged spatially or planarly, and a computer program. More specifically, the present invention relates to a data processing method for extracting a state change point from a steady state of an object to be measured or visualizing a state change state using a measured signal, and a computer program.

  Conventionally, signals measured in time series at multiple observation points arranged spatially or planarly from an object to be analyzed such as a living body to be analyzed or a system in a factory are analyzed, and the state in the object to be measured There are means for extracting points (parts, positions, etc.) where changes have occurred, and means for analyzing the relationship between the points. For example, using cross-correlation analysis or regression analysis, membrane potential imaging data, functional magnetic resonance image data, electroencephalogram, electromyogram, near-infrared spectroscopy, etc. measured in time series from any part of a living organ Means for analyzing and extracting the activation site in the measurement object, the relationship between each site, and the like are known.

The cross-correlation analysis is generally performed by the following procedure. For example, the object to be measured is captured in two dimensions, and the observation coordinates (observation part, observation position, etc.) on the two-dimensional object to be measured are (i, j), and the data measured in time series at the coordinates is represented by x _t The correlation coefficient is defined by the following equation, where y _t is the time-series reference data for examining the correlation with the measurement data x _t .

The numerator is a bivariate covariance, and the denominator is the standard deviation of each variable. Using the equation, for example, the reference data y _t, the coordinates on the object to be measured captured in two dimensions (i', j') When measured data in time series in, captured by the two-dimensional object Evaluation of the correlation between a point (part, position, etc.) specified by coordinates (i, j) on a measurement object and a point (part, position, etc.) specified by coordinates (i ′, j ′). it can. Or, for example, reference to data y _t, the input or when the time series signal data related to the output from the object to be measured, the measured strong measurement data x _t correlated with reference data y _t is measured to the object to be measured By specifying the coordinates (i, j) on the object, it is possible to specify a point that correlates with the signal related to the input / output (a portion where the state change is affected by the input signal / output signal).

In addition, by using regression analysis, it is possible to search for changes in the state of points (parts, positions, etc.) at which data is measured in time series. Specifically, a whitening method using an autoregressive model can be used. The autoregressive model is obtained by expressing a measured value η _t at a certain time t with a linear combination and a constant β using p past measured values, and is defined by the following equation.

ε _t is a prediction error, and its variance is σ _t ² . When the observation point (part, position, etc.) is in a steady state (when no state change occurs), the parameter vector δ = {α ₁ , α ₂ , α ₃ ,... Α _p , β, σ _t ² } Can be used to identify the steady-state linear dynamic characteristics of the observation point (part, position, etc.). The auto-regression model specified by the equation of “Equation 2” is applied to the measurement data x _t , and the residual e _t that is the difference between the calculated estimated value and the actually measured value is defined by the following equation.

Time zone deviates from the normal distribution in the time-series data e _t of the residual is, when the linear kinetics and different characteristics chronological x hours that occurred _t series eta _t, i.e. the observation point (site, location, etc. ) Is detected as the time zone when the state change occurred. When a set of residual data calculated in time series is used to extract state changes, the residual data is called “innovation”.

Although the data processing method for the univariate time series has been described above, the data can be analyzed in the same manner for the multivariate time series. Specifically, assuming that the number of variables is l, a vector X _t = (x _t ¹ , x _t ² ,... X _t ^l ) is represented, and its coefficient matrix is A _{i and the} intercept vector is β. , the time-series data e _t of the residual can be in the same manner as the case of time series of univariate, defined by the following equation.

The above can be applied to an exogenous variable autoregressive analysis in consideration of the influence of observation points (parts, positions, etc.) from the surroundings. As an exogenous variable type autoregressive analysis, for example, a method such as NEAREST-NEIGBORS AUTOREGRESIVE MODEL EXTERNAL INPUTS (NNARX) is known as an autoregressive model that considers the influence of observation points (parts, positions, etc.) only from the vicinity. It has been. This is done on the assumption that the observation points (parts, positions, etc.) are on a grid with one time series and are affected only by the time series in the vicinity thereof. Specifically, first, the Laplacian matrix L is defined as follows.

I is an l × l unit matrix, N is an l × l matrix, and an arbitrary point m affects the point l with respect to the point l indicating the observation point (part, position, etc.) on the grid. In the case of a neighboring point, “N _lm = 1” is defined, and in the case of other points, “N _lm = 0” is defined. k is the number of neighboring grids to consider. The residual data e _t ′ of X _t calculated using the Laplacian matrix L can be defined as follows (NNARX is disclosed in detail in Non-Patent Documents 1 to 3, etc.) Therefore, detailed description thereof is omitted here.)

Conventionally, using techniques such as those described above, analysis of signals measured in time series at observation points arranged spatially or planarly from a living body or a system in a factory has been performed. For example, an electroencephalogram multidimensional correlation analysis method as disclosed in Patent Document 1 is disclosed.
JP 7-51240 A Bosch-Bayard, J., Riera-Diaz, J., Biscay-Lirio, R., Wong, K.F.K., Galka, A., Yamashita, O., Sadato, N. Kawashima. Vazquez, E., Rodriguez-Rojas, R., Valdes-Sosa, P., Miwakeichi, F., and Ozaki to T. 2008 Galka, A., Ozaki, T., Bosch Bayard, J., Yamashita, and O., Whitening as a Tool for Estimating 12 in V. Spatialtemporal. -1315, 2006 Katanada, K., Matsuda, Y., and Sugishita, M., A Spatial-temporal Regression Model for the Analysis of Functional MRI Data14, Neuro14.

  However, the conventional analysis method has not established a method for extracting and visualizing the change of the dynamic state in multivariate time series data having a planar and spatial arrangement on the grid. In other words, intuitively understand how the point (part, position, etc.) where the state change occurred in the measurement target object (biological body, factory system, etc.) changes along the time axis. Was difficult.

  In addition, it is possible to represent the points (parts, positions, etc.) where the state change occurred and the relationship between each point as a statistic by analysis using conventional cross-correlation analysis or regression analysis. Thus, it is not possible to evaluate the dynamic properties of the data. Furthermore, the analysis by these methods requires that reference points and reference functions be defined, and there is an inconvenience that selection remains optional, and data reliability is somewhat lacking.

  Therefore, an object of the present invention is to provide an imaging data processing method capable of solving the above inconveniences, an imaging data processing system including a computer program, and the like.

  As means for solving the above problems, the following inventions are provided.

  In the first invention, a measurement data acquisition unit for acquiring time-varying imaging data recorded by associating a time change of a signal emitted from each point of the measurement object in a stationary state with the imaging data of the measurement object; and a measurement data acquisition unit The time-varying imaging data acquired by the unit is divided into analysis units divided into multiple regions, the analysis unit change data acquisition unit acquires change data for each divided analysis unit, and the analysis unit for each analysis unit A steady state data extraction unit that extracts steady state data from the change data acquired by the change data acquisition unit, a model function generation unit that generates a steady state model function for each analysis unit based on the extracted steady state data, and Using a steady-state model function, there is a residual data calculator that calculates residual data with change data in time series for each analysis unit Providing imaging data processing system having a state change method for extracting the imaging data.

  The second invention is based on the first invention and is further calculated by the residual data calculation unit based on a plurality of sets of time-varying imaging data obtained by repeated measurement acquired by the measurement data acquisition unit. Statistical analysis of the difference between the steady-state residual data and the residual data of other residual detection time zones for each section of time-varying imaging data partitioned into multiple regions using a set of residual data An imaging data processing system having an imaging data state change extraction method having a test value calculation unit for calculating test values in time series is provided.

  The third invention is based on the second invention, and further includes an imaging data state change extraction method having a test value mapping unit that maps the test values calculated in time series onto the imaging data of the object to be measured. An imaging data processing system is provided.

  In the fourth invention, a measurement data acquisition step for acquiring time-varying imaging data recorded by associating a time change of a signal emitted from each point of the measurement object in a stationary state with the imaging data of the measurement object; and a measurement data acquisition step A step for dividing the time-varying imaging data acquired in step 1 into analysis units divided into multiple regions, an analysis unit change data acquisition step for obtaining change data for each divided analysis unit, and an analysis unit for each analysis unit A steady state data extraction step for extracting steady state data from the change data acquired by the change data acquisition step, a model function generation step for generating a steady state model function for each analysis unit based on the extracted steady state data, and Residual data with change data for each analysis unit using steady state model functions It provides an operating method of an imaging data processing system having a state change detection method for imaging data with residual data calculating step of calculating the time series.

  The fifth invention is based on the fourth invention and is further calculated in the residual data calculation step based on a plurality of sets of time-varying imaging data obtained by repeated measurement acquired in the measurement data acquisition step. Statistical analysis of the difference between the steady-state residual data and the residual data of other residual detection time zones for each section of time-varying imaging data partitioned into multiple regions using a set of residual data And a method of operating an imaging data processing system having a method for extracting changes in state of imaging data having a test value calculation step for calculating test values in time series.

  According to a sixth aspect of the invention, there is provided an imaging data state change extraction method based on the fifth aspect, further comprising a test value mapping step of mapping the test value calculated in time series onto the imaging data of the object to be measured. A method for operating an imaging data processing system is provided.

  By the data processing method using the imaging data processing system of the present invention, it is possible to extract points (parts, positions, etc.) where a state change has occurred in an object to be measured (biological body, factory system, etc.) in time series. . In addition, it becomes possible to intuitively understand how the point (part, position, etc.) where the state change occurs changes.

  Further, according to the present invention, unlike the prior art, since the arbitrarily determined reference point is not used, the reliability of the state change point extraction is increased. Furthermore, since data processing can be performed using a plurality of data measured a plurality of times, the reliability can be increased as the number of measurements is increased.

  Furthermore, since the present invention can be applied to any data measured in time series from observation points arranged spatially or planarly from the object to be measured, it is expected to be used in various applications.

  Further, according to the present invention, a statistical test is performed on all the residuals calculated for each measurement without using the averaging method, so that the change in the dynamic state can be evaluated as the reliability.

  Furthermore, the present invention can detect the biological activity as a dynamic state change point on the time axis of the data even if the data cannot detect the biological activity based on the amplitude information.

Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings. In addition, this invention is not limited to these embodiment at all, and can be implemented with various aspects in the range which does not deviate from the summary.
<Outline of this embodiment>

In the data processing method using the imaging data processing system of the present invention, an arbitrary event (eg, breathing, external stress, explosive sound, etc.) occurs from an object to be measured (eg, heart, brain, factory system, etc.). When data obtained by measuring time-varying imaging data (eg, membrane potential imaging data, thermography, etc.) at least once is acquired, each point (part, position, etc.) of the object to be measured is used by using the data. State changes can be extracted in time series. In addition, as shown in FIG. 1, the mapping is displayed so that the state change of each point (part, position, etc.) of the measured object caused by an arbitrary event can be intuitively grasped. And
<Functional configuration of this embodiment>

  An example of functional blocks of the imaging data processing system of this embodiment is shown in FIG. As shown in FIG. 2, the “imaging data processing system” (0200) of the present embodiment includes a “measurement data acquisition unit” (0201), a “division unit” (0202), and an “analysis unit change data acquisition unit”. (0203), “steady state data extraction unit” (0204), “model function generation unit” (0205), and “residual data calculation unit” (0206). Moreover, you may have a "test value calculation part" (0207). Furthermore, you may have a "test value mapping part" (0208).

  Here, the functional blocks of the apparatus may be realized as hardware, software, or both hardware and software. Specifically, if a computer is used, a CPU, a RAM, a bus, or a secondary storage device (a storage medium such as a hard disk, a non-volatile memory, a CD-ROM or a DVD-ROM, and a read drive for the medium) ), Hardware components such as printing devices, display devices, and other external peripheral devices, I / O ports for the external peripheral devices, driver programs for controlling these hardware, other application programs, and information input User interface.

  In addition, these hardware and software processes the programs developed on the RAM with the CPU, and processes, stores, and outputs data stored on the memory and hard disk, and data input via the interface. Or used to control each hardware component. The present invention can be realized not only as an apparatus but also as a method. A part of the invention can be configured as software. Further, a software product used for causing a computer to execute such software, and a storage medium in which the product is fixed to a storage medium are naturally included in the technical scope of the present invention (the same applies throughout this specification). Is).

  Hereinafter, the “measurement data acquisition unit” (0201), “dividing unit” (0202), “analysis unit change data acquisition unit” (0203) of the “imaging data processing system” (0202) of the present embodiment, “Steady state data extraction unit” (0204), “Model function generation unit” (0205), “Residual data calculation unit” (0206), “Test value calculation unit” (0207), “Test value mapping” The functional configuration of the “part” (0208) will be described in detail.

  The “measurement data acquisition unit” (0201) is configured to acquire time-varying imaging data recorded by associating a temporal change of a signal emitted from each point of the measurement object in a stationary state with the imaging data of the measurement object. Yes. Here, the “measurement object” can measure signals in time series at observation points arranged in space or in a plane, such as a system in a factory, in addition to an organ in a living body such as a brain, heart, or lung. Everything is applicable. “Each point” of an object to be measured refers to a part located in each section when an object to be measured such as a brain or heart is arbitrarily or planarly or spatially partitioned on a grid, or a system such as a system in a factory. This corresponds to the position of each section when the measurement object is arbitrarily divided on the grid in a planar or spatial manner. The “measurement object in a stationary state” has the same meaning as the measurement object placed in a state where time-varying imaging data can be measured. That is, when the object to be measured is a living organ such as a brain or a heart, it means a state in which the object to be measured is placed or fixed under the measuring apparatus. Further, in the case of a system in a factory, it means a state in which the arrangement of devices in the system is fixed. `` Signal emitted from each point of the object to be measured '' refers to any data that can be measured in time series from the object to be measured, for example, electrical signals such as membrane potential, myoelectric potential, and electroencephalogram that can be measured from a living body, This applies to the temperature of the system in the factory. “Time-varying imaging data” means that the object to be measured as described above is partitioned on a planar or spatial grid on the imaging data (an image of the object to be measured in an arbitrary composition) and corresponds to each section This is data recorded by associating time-change data of a signal measured from a point (part, position, etc.) on an object to be measured in association with each section on imaging data.

  Here, as the means for measuring the “time-varying imaging data”, the following means can be considered. For example, when a biological organ or the like is used as a measurement object, a means for photographing a light emitted from each part of the measurement object with a CCD camera as a plurality of still images using a membrane potential sensitive dye can be considered. The membrane potential sensitive dye is a dye having a characteristic that the absorbance and the fluorescence intensity change in accordance with the change in the membrane potential at the point (part) where the dye is applied. That is, a plurality of still images of the object to be measured coated with the dye are photographed every predetermined time with a CCD camera, and the change in the amount of light of the dye applied to each point (part) of the object to be measured is recorded. It is possible to record a time change (a time change of membrane potential) of a signal emitted from each point (part) of the measurement object. As such a membrane potential sensitive dye, there are various types such as a stylyl compound, a cyanine compound, an oxonol compound, and a rhodamine derivative, and these can be used properly in consideration of a measurement site and measurement conditions. In such a case, each point is easily generated by using a still image of the measurement object photographed by the CCD camera as imaging data of the measurement object that correlates a time change of a signal emitted from each point of the measurement object. The time change of the signal can be recorded in association with the imaging data of the object to be measured.

  As another means for measuring time-varying imaging data when a biological organ or the like is used as an object to be measured, for example, first, imaging data of the object to be measured is prepared, and the object to be measured is planar or spatial on the imaging data. Partition on any arbitrary grid. Then, an electroencephalogram, an electromyogram, or the like is measured using an electrode or the like for each part of the object to be measured corresponding to each section. Then, the measured data may be recorded in association with each section of the imaging data.

  The means using the membrane potential sensitive dye has a demerit that the noise component increases as compared with the means for measuring the potential change for each part using the electrode as described above. It has the merit that the potential change of a plurality of parts can be measured simultaneously. The means for measuring the potential change for each part using one electrode or the like has the merit that the noise component is smaller than the means using the membrane potential sensitive dye, but on the other hand, the means for measuring a plurality of parts of the object to be measured. There is a demerit that it is difficult to measure potential changes at the same time. Therefore, it is desirable to use the measuring means properly according to the type of the object to be measured, the number of areas to be partitioned, and the like.

  In addition, as a means for measuring time-varying imaging data when a system in a factory or the like is an object to be measured, temperature change of the system in the factory may be measured using, for example, thermography. Even in such a case, by using a still image of the measurement object photographed by thermography, it is possible to easily record the time change of the signal generated by each point in association with the imaging data of the measurement object.

  The measuring means is an example, and the time-varying imaging data of the present invention may be time-varying imaging data measured from the object to be measured by other means.

  The measurement data acquisition unit (0201) acquires data obtained by measuring the time-varying imaging data as described above once or more. The number of measurements here is the time change of the signal generated by each point (part, position, etc.) of the measured object when an arbitrary event (eg, breathing, external stress, electrical stimulation, explosion sound, etc.) occurs once. Is measured once. Specifically, for example, it is assumed that a time change of a signal generated by each part to be analyzed in the brain when one breath (an action of inhaling once and exhaling) occurs is regarded as one time. The number of measurements of data acquired by the measurement data acquisition unit (0201) is not particularly limited, and may be, for example, 1, 2, 10, 30, 100, or more. However, considering the statistical test of the “test value calculation unit” (0207) described below, it is desirable that the number is approximately 30 or more. The data measured a plurality of times may be “data measured every time” or “data measured continuously a plurality of times”. “Data measured every time” means “measurement start → arbitrary circumstances → measurement end” → “measurement start → arbitrary event → measurement end” → “...”. For example, when measuring change imaging data. In such a case, the time-varying imaging data is classified and stored for each measurement time. On the other hand, “data measured continuously multiple times” means “measurement start → any event → any event → ... end of measurement”, etc. It is. In such a case, the time-varying imaging data is stored without being classified for each measurement time.

  Here, the means by which the measurement data acquisition unit (0201) acquires the time-varying imaging data as described above is not particularly limited, and for example, by extracting the time-varying imaging data stored in the internal memory of the same device. It may be acquired, or may be acquired by wired / wireless communication with an external device that stores time-varying imaging data.

  The “dividing unit” (0202) is configured to divide the time-varying imaging data acquired by the measurement data acquiring unit into analysis units divided into a plurality of regions. “Analysis unit” is the smallest unit for performing analysis such as state change extraction using time-varying imaging data measured from the object under measurement, and divides the image data of the object under measurement into an arbitrary number of areas. The time-varying imaging data measured in one section is used as one analysis unit. The means for the dividing unit (0202) to divide the acquired time-varying imaging data into analysis units is not particularly limited, but may be as follows.

  For example, when the time-varying imaging data is data obtained by capturing the object to be measured in two dimensions (eg, imaging data only on the surface of the object to be measured), a still image captured by a CCD camera or a thermography is captured for each pixel. You may implement | achieve by dividing. Then, the time-varying imaging data corresponding to each section (pixel) may be divided as one analysis unit. In this case, the dividing unit (0202) divides the temporal change imaging data into L × M (pieces) analysis units (sections) according to the number of pixels (L × M (pixels)).

  In addition, even when time-varying imaging data is data that captures the object to be measured in three dimensions (eg, a still image of the surface of the sample obtained by slicing the object to be measured), a CCD camera or thermography This can be achieved by dividing the captured still image into pixels. In such a case, the dividing unit (0202) also divides the time-varying imaging data into the sliced (N sliced) samples, and divides them into L × M × N (units) analysis units (sections). To divide.

  In addition, when the time-varying imaging data is obtained by measuring a signal (potential change, etc.) for each point (part, position, etc.) on the object to be measured corresponding to each section of the imaging data using electrodes etc. The time-varying imaging data is stored as data for each section. Therefore, in such a case, the dividing unit (0202) can divide the time-varying imaging data of one section as one analysis unit according to the section.

The division means is merely an example, and the time-varying imaging data may be divided into analysis units by other means. For example, instead of using one pixel as one analysis unit, time-varying imaging data of a plurality of pixels (for example, p ² pixels adjacent to form a rectangle: p is a natural number) is one analysis unit. It is good.

  Here, when the time change imaging data acquired by the measurement data acquisition unit (0201) is the time change imaging data measured a plurality of times, the division unit (0202) divides the time change imaging data for each measurement time. You may comprise. The dividing means may be as follows.

  For example, when the time-varying imaging data measured a plurality of times is “data continuously measured a plurality of times” (“measurement start → any event → any event → ・ ・ ・ end of measurement”), The measurement time in each measurement is determined (for example, 5 sec / time), and time-varying imaging data is measured. Then, the dividing unit (0202) identifies the time length of the acquired time-varying imaging data, and calculates the number of times of measurement (R times) using the measurement time (eg, 5 sec / time) stored in advance. May be. The dividing unit (0202) may be configured to divide the acquired time-varying imaging data into R groups at predetermined time (measurement time) intervals from the beginning.

  In addition, time-varying imaging data measured multiple times is "data measured every time" ("measurement start-> arbitrary circumstances-> measurement end"-> "measurement start-> arbitrary event-> measurement end"-> "... In the case of ()), since the time-varying imaging data has already been divided for each measurement time, the dividing unit (0202) does not need to divide the time-change imaging data measured a plurality of times for each measurement time.

  When acquiring the time-varying imaging data measured a plurality of times by the measurement data acquiring unit (0201), the dividing unit (0202) uses the time-varying imaging data of one section of one measurement as one analysis unit. To divide. That is, when the measurement data acquisition unit (0201) captures the object to be measured in three dimensions and acquires time-varying imaging data measured R times, it is divided into R × L × M × N analysis units. It becomes.

  The “analysis unit change data acquisition unit” (0203) is configured to acquire change data for each divided analysis unit. “Change data” is data relating to temporal changes in signals (electrical signals and the like) measured from each point (part, position, etc.) of the object to be measured. That is, when the time-change imaging data is obtained by measuring a change in the amount of light at each part of a living organ using a membrane potential sensitive dye, the change data corresponds to data indicating a change in the amount of light at each part. In addition, when the time-change imaging data is obtained by measuring the potential of each part of a living organ using an electrode or the like, the change data corresponds to the potential change of each part. Further, when the time change imaging data is obtained by measuring the temperature change of the system in the factory using thermography, the data indicating the temperature change at each position is applicable. Here, when one analysis unit is specified by a plurality of pixels, the change data may be an average value of a plurality of change data acquired from the plurality of pixels.

  Note that the number of pieces of change data acquired from one analysis unit varies depending on measurement conditions. That is, for example, in the case of time-varying imaging data in which 256 still images are recorded every 0.02 sec in one measurement, 256 pieces of change data are acquired from one analysis unit. Similarly, when the potential is measured with an electrode or the like, change data corresponding to the number of data recorded in one measurement is acquired.

  The “steady state data extraction unit” (0204) is configured to extract steady state data from the change data acquired by the analysis unit change data acquisition unit for each analysis unit. The “steady state data” is change data when the object to be measured is in a steady state (a state that is not affected when an arbitrary event occurs). Means for identifying and extracting the steady state data from the change data may be as follows. For example, time-varying imaging data measured at a time is measured by taking a sufficient time (eg, 3 sec) before and after an arbitrary event occurs. Then, the steady state data extraction unit (0204) uses the change data for the first predetermined time (eg, 0.52 sec) as the steady state data from the change data acquired by the analysis unit change data acquisition unit (0203). You may comprise so that it may acquire.

  As another means, for example, time-varying imaging data measured at one time is not affected by the occurrence of any event, rather than being affected by the occurrence of any event. Take a long enough to measure. Then, when the average value of the change data is calculated, the steady state extraction unit (0204) specifies the change data that deviates most from the average value, and sets a predetermined time width (arbitrarily set in advance) centered on the data. You may comprise so that change data other than the data contained may be acquired as steady state data.

  The “model function generation unit” (0205) is configured to generate a steady state model function for each analysis unit based on the extracted steady state data. A “steady state model function” is a dynamic model expression such as an autoregressive model, an autoregressive moving average model, or a state space model, and is calculated by a statistical method such as a least squares method or a maximum likelihood method using a Kalman filter. . The steady state model function calculated for each analysis unit may be a univariate model formula calculated using only change data of one section (part, position, etc.) specified by the analysis unit. Alternatively, it may be a multivariate exogenous variable type model formula calculated in consideration of the influence of surrounding sections (parts, positions, etc.) of one section specified by the analysis unit. Since these model formulas can be calculated using the prior art, a detailed description thereof is omitted here. When calculating the exogenous variable type model formula, NEEAREST-NEIGBORS AUTOREGRESIVE considering the effect of only the section (part, position, etc.) in the vicinity of one section (part, position, etc.) specified in the analysis unit. A multivariate model equation may be calculated using a technique such as MODEL WITH EXTERNAL INPUTS (NNARX). The details of the method are described in Non-Patent Documents 1 to 3, and thus detailed description thereof is omitted here.

  The model function generation unit (0205) generates a steady state model function for each analysis unit. That is, for example, when the analysis unit divided by the dividing unit (0202) is R × L × M × N (pieces), the model function generating unit (0205) is R × L × M × N (pieces) stationary. A state model function will be generated.

  The “residual data calculation unit” (0206) is configured to calculate residual data with change data in time series for each analysis unit by using the generated steady state model function. That is, the residual data calculation unit (0206) calculates residual data that is a difference between the estimated value and the actual measurement value for each analysis unit, using the steady state model function generated for each analysis unit.

For example, assume that T data is recorded in one measurement (eg, T still images are recorded in one measurement), and in the r-th measurement (1 ≦ r ≦ R | r, R: natural number). Sections (1 ≦ l ≦ L, 1 ≦ m ≦ M, 1 ≦ n ≦ N | 1, m, n, L, M, N: natural numbers specified by (l, m, n) on a three-dimensional grid ), The t-th (1 ≦ t ≦ T | t, T: natural number) change data from the beginning of the actually measured change data is ^defined as x _t ^lmnr (r, R, Unless otherwise specified, l, L, m, M, n, N, t, and T are the same as the above assumptions). In such a case, using the NNARX method, the section specified by “(l, m, n) is (l ₁ ≦ l ≦ l ₂ , m ₁ ≦ m ≦ m ₂ , n ₁ ≦ n ≦ n ₂ | l ₁ , l ₂ , m ₁ , m ₂ , n ₁ , n ₂ : natural number) are assumed to be affected only by the section, and the r-th time calculated using the steady state model function The residual data e _t ^lmnr of the analysis unit specified by (l, m, n) can be calculated by the following equation.

  The residual data calculation unit (0206) uses, for example, the calculation formula shown in Equation 7 to calculate the estimated value calculated using the steady state model function and the actually measured change data for each analysis unit. Residual data that is the difference is calculated in time series. Here, assuming the above, the number of residual data calculated by the residual data calculation unit (0206) is R × L × M × N × T (pieces) of residual data.

  The calculated residual data may be displayed as a mapping as shown in FIG. FIG. 3 is a mapping of R × T (pieces) residual data calculated in a certain section (l, m, n). The vertical axis represents the number of measurements, and corresponds to r times. Although the horizontal axis indicates time, when T pieces of data are recorded at equal time intervals (p (sec)), the horizontal axis can correspond to the order t in which the data is recorded. In other words, FIG. 3 shows the magnitude of the residual data calculated in a certain section (l, m, n) by color shading along the measurement time axis for each measurement count. In the figure, the white portion is a portion with a large residual, and the dark portion indicates a portion with a small residual. From the figure, in the section (part, position, etc.) specified by (l, m, n) of the object to be measured, a large residual is calculated between 1.0 sec and 2.0 sec from the start of measurement, It can be read that this is the same tendency in a plurality of measurements. That is, from the figure, the section (part, position, etc.) specified by (l, m, n) of the object to be measured is relatively relatively between 1.0 sec and 2.0 sec from the start of measurement due to some factor. It can be intuitively understood that a large state change (potential change) has occurred.

  The “test value calculation unit” (0207) is a plurality of sets of residual data calculated by the residual data calculation unit based on the multiple sets of time-varying imaging data obtained by repeated measurement acquired by the measurement data acquisition unit. Using the difference data, statistically test the magnitude of the difference between the residual data in the steady-state residual data and the residual data in other residual detection time zones for each section of the time-varying imaging data partitioned into multiple regions. It is configured to calculate values in time series.

  Here, the “number of sets” of time-varying imaging data is generated by each point (part, position, etc.) of the measured object when an arbitrary event (eg, breathing, external stress, explosion sound, etc.) occurs once. Time-varying imaging data in which signal temporal changes are recorded is considered as a set of time-varying imaging data. That is, “a plurality of sets of time-varying imaging data obtained by repeated measurement” refers to time-varying imaging data obtained by repeatedly measuring the above. This assumption is the same for all “number of sets” below.

  “Multiple sets of residual data” means residual data calculated for each measurement (for each set) using time-varying imaging data for multiple measurements (multiple sets of time-varying imaging data) Is applicable. Specifically, a steady state model function is generated using time change data (one set of time change imaging data) of one measurement, and one set of residual data is generated using the steady state model function. Is calculated. Then, the process is performed on all time-varying imaging data for a plurality of measurements, and a plurality of sets of residual data are calculated.

The test value calculation unit (0207) groups a plurality of sets of residual data for each section of time-varying imaging data partitioned into a plurality of areas, and detects the residual data in the steady state of the section and other residuals. Statistically test the magnitude of the difference in the residual data for the time zone. The details will be described below.
<Statistical test of test value calculation part>
<< Extract "Residual data in steady state" and "Residual data in other residual detection time zones" for each section >>

First, the test value calculation unit (0207) calculates the residual data (R × T (pieces) of the section specified by (l, m, n) from the residual data of all sets (measured R times). )). Then, steady state residual data A ^lmn = {a ₁ , a ₂ ,..., A _i } is extracted from the extracted R × T (pieces) residual data. The steady state specifying means can be specified by means similar to the above-described “steady state data extraction unit” (0204). Specifically, the test value calculation unit (0207) may extract residual data for a predetermined time (eg, 0.52 sec) from the start of measurement as residual data in a steady state. This corresponds to the residual data of the portion surrounded by the frame A in FIG. Next, the test value calculation unit (0207) extracts residual data B _h ^lmn = {b ₁ , b ₂ ,..., B _j } in other residual detection time zones. The “other residual detection time zone” is an arbitrary time zone within one measurement time, and corresponds to the residual data of the portion surrounded by the frame B in FIG. Here, in FIG. 3, the position of the frame of B is merely an example, and the test value calculation unit (0207) slides the frame of B sequentially along the time axis from the measurement start time to the measurement end time, The “other residual detection time zones” are identified in order. Then, residual data in the specified time zone is extracted. At this time, another residual detection time zone (frame B in FIG. 3) may overlap with the steady-state time zone (frame A in FIG. 3). The time width of “other residual detection time zone” (the width on the time axis of the frame B in FIG. 3) is an arbitrary design matter. However, it is preferable to set the time width as small as possible.
<< Statistical test of difference between representative value of "residual data in steady state" and representative value of "residual data in other residual detection time period">>

Next, the test value calculation unit (0207) calculates the residual values of the residual data A ^lmn = {a ₁ , a ₂ ,..., A _i } in the steady state and other residual detection time zones. In order to examine whether the magnitude of the difference between the representative values of the data B _h ^lmn = {b ₁ , b ₂ ,..., B _j } has a significance of a predetermined significance level or higher, a statistical test is performed. As a statistical test method, for example, a parametric test such as a t test can be used. Hereinafter, a statistical test by the t test will be briefly described.

First, in the case of t-test, an average value of each residual data can be applied as a representative value of each residual data. Specifically, the representative value A ^lmn ′ of the residual data A ^lmn is calculated by (a ₁ + a ₂ +... + A _i ) / i. The representative value B _h ^lmn ′ of the residual data B _h ^lmn is calculated by (b ₁ + b ₂ +... + B _j ) / j. Then, ^{assuming that} the respective non-uniform variances of the residual data A ^lmn and B _h ^lmn are U _A ^lmn and U _Bh ^lmn , the test value t _h ^lmn is calculated by the following equation.

The test value calculation unit (0207) calculates and stores the test value t _h ^lmn of each time zone while sequentially specifying “other residual detection time zones” along the time axis. When the test values t _h ^lmn of all “other residual detection time zones” are calculated, other sections (eg, (l, m, n + 1)) are specified, and the test values t _h ^{lmn + are} similarly determined. ¹ is calculated. Thereafter, the calculation processing of the test value t _h ^lmn as described above is performed for all the sections.

Then, when the test value calculation unit (0207) performs the calculation process of the test value t _h ^lmn of all the sections, only the test values having a predetermined significance level or higher are extracted from the calculation values, and the test value sections ( l, m, n) and information specifying the time zone from the start of measurement are stored so as to be output to the user. By referring to the information, the user can identify a point (a part, a position, etc.) where a state change having a certain level of significance or more has occurred in the object to be measured and a time zone. It is also possible to detect the magnitude of the state change. The “time zone from the start of measurement” associated with the extracted test value is a case where T data is recorded at equal time intervals (p (sec)) in one time-varying imaging data measurement. The time zone from the start of measurement of the change data recorded in the tth position can be specified by p × t (sec).

  Here, the statistical test based on the t-test has been described above. However, a non-parametric test such as a surrogate method can also be used. A condition for properly using these tests is, for example, “the number of residual data in steady state and the number of residual data in other residual detection time zones are both 30 or more, and their distributions are both normal. If a distribution can be considered, a parametric test is used. In other cases, a non-parametric test is used. " However, the above conditions are just general guidelines, and it is not always necessary to select a test method according to the conditions. Since the surrogate method is a generally known technique, a detailed description thereof is omitted here.

The “test value mapping unit” (0208) is configured to map the test values calculated in time series on the imaging data of the object to be measured in time series. That is, when the test value mapping unit (0208) extracts only test values having significance equal to or higher than a predetermined significance level from the calculated test values, an interval (l, m, n) corresponding to the test value is extracted. And the time zone (p × t) from the start of measurement. Then, for each time zone from the start of measurement, the extracted test value is mapped to each section on the imaging data of the object to be measured. As a mapping method, a color plot or a contour plot can be considered. FIG. 4 shows an example of a diagram in which test values of a predetermined significance level or higher are mapped onto the imaging data of the object to be measured in time series (every time zone (p × t) from the start of measurement). In the figure, the upper four figures are diagrams in which test values having significance of a predetermined significance level or higher are mapped on the imaging data of the object to be measured. The figure shows the mapping in any four time zones. Of course, it is possible to create a mapping diagram in other time zones by the same method. The lower side of the figure is data relating to any event that has occurred on the device under test. As shown in FIG. 4, by displaying the time zone with an arbitrary event together, it is possible to intuitively grasp the transition of the state change point in the measured object corresponding to the arbitrary event. In addition, the mapping diagram shown on the upper side in FIG. 4 is not only shown as shown in the figure, but may be displayed in a sliding manner along the time axis. By displaying in this way, the transition of the state change point of the object to be measured can be grasped more intuitively. Note that the value of the predetermined significance level is not particularly limited, and can be arbitrarily set according to the measurement target, the experiment content, and the like. For reference, when testing an electrical signal emitted from a living body or the like, it is preferably 95% or more.
<Hardware configuration of this embodiment>

  FIG. 5 is a diagram illustrating an example of a configuration when the functional configuration is realized as hardware.

  Hereinafter, an example of means for realizing the present embodiment will be described with reference to the hardware diagram of FIG. As shown in the figure, the imaging data processing system of this embodiment includes a “measurement data acquisition unit”, a “division unit”, an “analysis unit change data acquisition unit”, a “steady state data extraction unit”, and a “model function generation unit”. , “Residual data calculation unit”, “test value calculation unit”, “test value mapping unit”, etc., “CPU” (0501), “main storage device” (0502), “program storage device” (0503) ), “Secondary storage device” (0504), “user I / F” (0505), “external device I / F” (0506), “display” (0507), “bus” (0508), etc. Yes.

  The main storage device (0502) is a storage device capable of dynamically rewriting data during program execution. The main storage device (0502) provides a work area such as a stack and a heap necessary for executing the program stored in the program storage device (0503). The main memory (0502) also includes time-varying imaging data acquired according to the measurement data acquisition program, change data acquired for each analysis unit from time-varying imaging data divided for each analysis unit according to the division program, and a steady state model. Grouped for each partition according to the steady state model function generated for each analysis unit according to the function generation program, residual data calculated for each analysis unit using the steady state model function according to the residual data calculation program, and statistical test program Data obtained from the test data calculated for each partition using the residual data grouped for each partition, or the detected value data calculated from the calculated detection value data. According to the test value data or the test value mapping program Or hold such mapped data the significance level or more test value data.

  The secondary storage device (0504) is a storage device that can dynamically rewrite data during program execution. Even if the signal processing device is turned off, stored data is not erased. The secondary storage device holds significance level data that defines a predetermined significance level in the statistical test, mapping data for mapping test values that are equal to or higher than the predetermined significance level, and the like.

  The user I / F (0505) receives a data processing instruction signal and the like from the user. The external device I / F (0506) communicates with the external device by wire or wireless, and receives time-varying imaging data measured from the object to be measured. The display (0507) displays an interface that guides the input of the user's instruction signal with characters, still images, moving images, and the like, or displays mapping data that maps processing data.

  Hereinafter, the imaging data processing system of the present embodiment acquires a plurality of sets of time-varying imaging data obtained by repeated measurement, and extracts the state change point of the measurement object with high accuracy using the data. An example of processing for mapping the results will be described.

  First, when an instruction signal for starting data processing is received via the user I / F (0505), the external device I / F (0506) is executed in accordance with the measurement data acquisition program developed in the work area of the main storage device (0502). ) To obtain time-varying imaging data, which is data measured from the object to be measured, and store it in the data area of the main memory (0502). At this time, although not shown in FIG. 5, the system may include an HDD or the like, and the time-varying imaging data stored therein may be acquired.

  Next, the acquired time-varying imaging data is divided for each analysis unit in accordance with the division program developed in the work area of the main storage device (0502). Specifically, first, time-varying imaging data is divided into R groups for each number of measurements. Then, the object to be measured is partitioned into a plurality of regions (L × M × N) and divided into a total of R × L × M × N analysis units. After that, according to the steady state model function generation program developed in the work area of the main storage device (0502), when time series change data is obtained for each analysis unit, steady state change data of a predetermined time width is extracted from the time series change data. To do. Then, using the extracted steady-state change data, an autoregressive model (steady-state model function) is generated for each analysis unit using NNARX or the like, and stored in the data area of the main memory (0502).

  Next, according to the residual data calculation program developed in the work area of the main storage device (0502), the time series residual data is calculated for each analysis unit using the steady state model function generated for each analysis unit. To do. Then, the residual data calculated for each analysis unit is grouped for each partition (L × M × N groups) and stored in the data area of the main storage device (0502) as residual data for each partition.

  Next, according to a statistical test program developed in the work area of the main memory (0502), a statistical test of residual data is performed for each partition. Specifically, first, when the residual data for each section of one section is taken out, the residual data in the steady state is extracted therefrom. Next, residual data is extracted by specifying other residual detection time zones, and respective representative values are calculated. The magnitude of the representative difference is statistically tested by a technique such as t-test to calculate a test value. Thereafter, other residual detection time zones are specified in order according to the time axis. When the test value calculation process is performed for all other residual detection time zones, the calculated time-series test values are stored in the data area of the main memory (0502) as test value data. And the said process is implemented with respect to all the residual data for every division, and the calculated data are stored as verification value data for every division. Thereafter, when the significance level data stored in the secondary storage device (0504) is taken out, only the test value data corresponding to the predetermined significance level or more specified by the significance level data is selected from the test value data for each section. Extracted and stored in the data area of the main memory (0502) as the significant level test value data for each section. The data may be displayed as numerical data on a display (0507) or the like according to a request from the user.

Next, when mapping data is taken out from the secondary storage device (0504) according to the test value mapping program developed in the work area of the main storage device (0502), the measured data included in the time-varying imaging data on the mapping data Expand the static image of the object, and use the significance level test value data for each compartment to show the size of the test value that is greater than or equal to the prescribed significance level in time series. To be identifiable. Then, the mapped data is output through the display (0507).
<Processing flow of this embodiment>

  An example of the processing flow of this embodiment is shown in the flowcharts of FIGS. FIG. 6 shows an example of processing for calculating residual data in time series for each analysis unit using the acquired time-varying imaging data. FIG. 7 shows an example of a process of extracting and mapping the state change points of the object to be measured in a time series with high accuracy using the acquired multiple sets of time-change imaging data. 8 to 10 are examples of a part of the processing flow of the present embodiment.

Note that the data processing method of the imaging data processing system of the present embodiment may be a simple method of acquiring data and processing data according to the following flow, or the following The flow may be realized as a series of operation methods of the electronic computer.
<< Flow of processing to calculate residual data in time series (Fig. 6) >>

  First, time-varying imaging data obtained by measuring an object to be measured is acquired from an external device or a built-in HDD (S0601). Thereafter, the time-change imaging data is divided into analysis units divided into a plurality of regions for each measurement time (S0602). At this time, when the number of times of measurement is one time, the time-varying imaging data is not divided into each measurement time, and is divided into analysis units divided into a plurality of regions.

Then, an autoregressive model (steady state model function) is generated using a technique such as NNARX using steady state data for each analysis unit (S0603). Thereafter, residual data is calculated for each analysis unit using the steady state model function generated for each analysis unit (S0604).
<< Flow of processing to extract and map state change points of DUT in time series (Fig. 7) >>

  First, a plurality of sets of time-varying imaging data, which is data obtained by repeatedly measuring an object to be measured a plurality of times, is acquired from an external device or a built-in HDD (S0701). Thereafter, the temporal change imaging data is divided into analysis units divided into a plurality of regions for each set (for each measurement time) (S0702).

  Then, an autoregressive model (steady state model function) is generated using a technique such as NNARX using steady state data for each analysis unit (S0703). After that, residual data is calculated for each analysis unit using the steady state model function generated for each analysis unit (S0704).

  Then, the residual data calculated for each analysis unit is grouped for each section (S0705), and the residual data is statistically tested for each section (S0706). Thereafter, the result of the statistical test is mapped and displayed in time series on the imaging data of the object to be measured (S0707).

Hereinafter, the details of steps S0703, S0704, and S0706 in the flow of reason will be described.
<< Generation of Steady State Model Function: Step S0703 (FIG. 8) >>

  First, when one analysis unit is specified (S0801), change data (for example, potential change, light amount change, temperature change) of the analysis unit is acquired (S0802). Then, steady state data is extracted from the list (S0803). Next, an autoregressive model (steady state model function) is generated using a method such as NNARX using the extracted steady state data (S0804).

Thereafter, the above processing (S0801 to S0804) is repeated until a steady state model function of all analysis units is generated (S0805).
<< Calculation of residual data: S0704 (FIG. 9) >>

  First, when one analysis unit is specified (S0901), one change data is extracted along the time axis (S0902). Then, an estimated value is calculated using the generated steady state model function (S0903), and residual data is calculated (S0904). The above processing (S0902-S0904) is repeated until residual data of all change data is calculated in time series (S0905).

Thereafter, the above processing (S0901 to S0905) is repeated until residual data of all analysis units is calculated (S0906).
<< Statistical test: S0706 (Fig. 10) >>

  First, when one section is specified (S1001), residual data in a steady state is extracted from the residual data of the section (S1002), and a representative value is calculated (S1003). Then, according to the time axis, one residual detection time zone is specified from the residual data of the section (S1004), and when the residual data of the residual detection time zone is extracted (S1005), the representative value is obtained. Calculate (S1004). Thereafter, a statistical test is performed on the difference between the representative value in the steady state and the representative value in one residual detection time zone (S1007), and the test value is stored (S1008). The processing (S1004 to S1008) is repeated until test values for all residual detection time zones are calculated (S1009).

  When the test values for all the residual detection time zones are calculated, only test values that are equal to or higher than a predetermined significance level are extracted (S1010).

Thereafter, the processing (S1001 to S1010) is repeated until it is performed on the residual data of all sections.
<Concrete examples using this embodiment>
<< Purpose of Examples >>

The left figure of FIG. 11 is a photograph of the rat brain as seen from the ventral side, and the right figure is a partial enlarged schematic view thereof. In the figure, it is known that a region related to respiration such as pFRG exists in a region surrounded by a square (1101), and an output signal is sent to the C4 spinal nerve and phrenic nerve. The purpose of this embodiment is to calculate how the activation state of the nerve cells in the area enclosed by the square (1101) changes according to respiration, and to intuitively map it so that it can be grasped intuitively. To do.
<< Time-varying imaging data >>

In this example, a membrane potential sensitive dye was used, and the change in the amount of light on the surface of the object to be measured (brain) when breathing occurred was recorded as a plurality of still images by a CCD camera. The still image was 100 × 100 pixels, and 256 points were recorded every 0.02 sec. In response to one breath (inhaled once and exhaled once), measurement was performed once (256 points), and the above was measured a total of 27 times. In one measurement, a sufficient time (about 1 sec) was secured before breathing occurred. In addition, the electrical signal was measured from C4, and synchronization was made using the rising edge of the signal so that the phases of respiration of the data measured multiple times were aligned. For reference, data measured from C4 is shown in FIG. In the figure, the upper side is measurement data, and the lower side is data after frequency analysis is performed on the upper data.
<< Division into analysis units >>

In this example, time-varying imaging data obtained by capturing the surface of the object to be measured in two dimensions was analyzed. That is, as analysis unit division processing, time-varying imaging data was first divided into 27 groups for each number of measurements, and the object to be measured on the still image was partitioned into L × M regions. Specifically, the area was divided into 100 × 100 areas according to the number of pixels of the still image. That is, in this embodiment, the time-varying imaging data is divided into 27 × 100 × 100 analysis units.
<< Generation of steady state model function >>

In this embodiment, the steady state model function is generated using the NNARX method. First, the neighborhood that considers the influence is only 8 pixels located in the vicinity, and the lag of the model is 3. Then, the steady state was set for 0.52 sec from the start of measurement.
<< Calculation of residual data >>

  Using the generated steady state model function, time series residual data was calculated for each analysis unit. 256 residual data were calculated for each analysis unit.

FIG. 3 shows a diagram in which the temporal change of the residual data of one section (pixel) is mapped. The vertical axis corresponds to the number of measurements, and the horizontal axis corresponds to the time axis. In the figure, 256 × 27 residual data are mapped. In the figure, a dark portion indicates that the residual is small, and a white portion indicates that the residual is large. From the figure, it can be seen that a large residual is calculated in approximately 1.0 sec to 2.0 sec in all measurements. That is, it can be seen that the state change occurs at the site of the object to be measured corresponding to the section (pixel) in about 1.0 sec to 2.0 sec.
<< Statistical test of residual >>

  In this embodiment, the steady state is set for 0.52 sec from the start of measurement, and the other residual detection time zones are specified in order of 0.02 sec in accordance with the time axis from the start of measurement. In FIG. 3, a portion surrounded by a square A corresponds to residual data in a steady state, and a portion surrounded by a square B corresponds to residual data in one other residual detection time zone. That is, the residual data in the steady state has 26 × 27 points of residual data, and the residual data in one residual detection time zone has 1 × 27 points of residual data.

The stationary data and the other residual data can be regarded as a normal distribution, and the number of data is approximately 30 or more, so the statistical test was performed using the t test. As the predetermined significance level, 95% was applied in accordance with industry practice.
<< Mapping of test values >>

  FIG. 4 shows how the test values are mapped in time series. In the figure, the upper side is obtained by mapping the test value of a predetermined significance level or higher on the imaging data of the object to be measured, and the lower side is a C4 signal plotted to show the timing of respiration. In the figure, the upper part is a plot of the portion where the residual having a magnitude of a predetermined significance level or more is calculated in each time zone. Note that the four time zones in the figure are merely examples, and similar mapping diagrams can be displayed in other time zones. Also, these mapping diagrams can be displayed in a slide display.

From FIG. 4, it is possible to easily grasp a site that is activated before the C4 signal reaches a peak, a site that is activated in synchronization, and a site that is activated with a delay of about 2.5 seconds.
<Effect of invention>

  According to the imaging data processing system and the data processing method of the present invention, the state change points can be extracted with high accuracy in time series on the imaging data of the measured object to be measured (such as a living body or a factory system). In addition, it becomes possible to intuitively understand how the state change point changes along the time axis.

The figure for demonstrating the outline | summary of this invention Functional block diagram of the present invention Example of mapping of residual data of one section An example of a diagram in which the test value is mapped onto the still image of the object to be measured The conceptual diagram which showed an example of the hardware constitutions of this invention 1 is a flowchart showing the flow of processing of the present invention. FIG. 2 is a flowchart showing the processing flow of the present invention. FIG. 3 is a flowchart showing the processing flow of the present invention. FIG. 4 is a flowchart showing the processing flow of the present invention. FIG. 5 is a flowchart showing the processing flow of the present invention. FIG. 1 for explaining a specific example using the present invention. FIG. 2 for explaining a specific example using the present invention.

Explanation of symbols

0201 Processing data acquisition unit 0202 Division unit 0203 Analysis unit change data acquisition unit 0204 Steady state data extraction unit 0205 Model function generation unit 0206 Residual data calculation unit 0207 Test value calculation unit 0208 Test value mapping unit

Claims (6)

A measurement data acquisition unit for acquiring time-varying imaging data recorded by associating a time change of a signal emitted from each point of the measurement object in a stationary state with the imaging data of the measurement object;
A division unit that divides the time-varying imaging data acquired by the measurement data acquisition unit into analysis units divided into a plurality of regions,
An analysis unit change data acquisition unit for acquiring change data for each divided analysis unit;
For each analysis unit, a steady state data extraction unit that extracts steady state data from the change data acquired by the analysis unit change data acquisition unit;
A model function generator that generates a steady state model function for each analysis unit based on the extracted steady state data;
An imaging data processing system having an imaging data state change extraction method including a residual data calculation unit that calculates residual data with change data in a time series for each analysis unit using a generated steady state model function.   Partitioning into multiple regions using multiple sets of residual data calculated by the residual data calculation unit based on multiple sets of time-varying imaging data obtained by repeated measurement acquired by the measurement data acquisition unit A test value calculator that statistically tests the magnitude of the difference between the residual data in the steady state and the residual data in other residual detection time zones and calculates the test values in time series An imaging data processing system having the imaging data state change extraction method according to claim 1.   The imaging data processing system having the imaging data state change extraction method according to claim 2, further comprising a test value mapping unit that maps the test values calculated in time series onto the imaging data of the object to be measured in time series. A measurement data acquisition step for acquiring time-varying imaging data recorded by associating a time change of a signal emitted from each point of the measurement object in a stationary state with the imaging data of the measurement object;
A division step of dividing the time-varying imaging data acquired in the measurement data acquisition step into analysis units divided into a plurality of regions;
An analysis unit change data acquisition step for acquiring change data for each divided analysis unit;
For each analysis unit, a steady state data extraction step for extracting steady state data from the change data acquired by the analysis unit change data acquisition step;
A model function generation step for generating a steady state model function for each analysis unit based on the extracted steady state data;
Method of operating an imaging data processing system having a state change extraction method of imaging data having a residual data calculation step for calculating residual data with change data in a time series for each analysis unit using the generated steady state model function .   Partitioning into multiple regions using multiple sets of residual data calculated in the residual data calculation step based on multiple sets of time-varying imaging data obtained by repeated measurement acquired in the measurement data acquisition step A test value calculation step that statistically tests the difference between the residual data in the steady state and the residual data in other residual detection time zones for each section of the time-varying imaging data, and calculates the test values in time series A method for operating an imaging data processing system having the imaging data state change extraction method according to claim 4.   6. The operation of the imaging data processing system having the imaging data state change extraction method according to claim 5, further comprising a test value mapping step of mapping the test values calculated in time series onto the imaging data of the object to be measured in time series. Method.