Activity identification of a brain region

By incorporating a GARCH process to model variable conditional variance in fMRI signals, the method enhances the accuracy of brain activity identification in fMRI signal processing, addressing the limitations of existing GLM-based approaches and improving the precision of brain activity mapping.

WO2024231716A1PCT designated stage expired Publication Date: 2024-11-14HAKIMDAVOODI HAMIDREZA +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/IB2023/054815
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-10
Publication Date
2024-11-14

AI Technical Summary

Technical Problem

Current fMRI signal processing methods, particularly those based on the general linear model (GLM), face challenges in accurately identifying brain activity due to varying conditional variance across different brain regions, leading to biased parameter estimates and incorrect statistical inference, especially when dealing with a small number of fMRI images and noise from sources like head motion and scanner noise.

Method used

The method employs a generalized autoregressive conditionally heteroscedastic (GARCH) process to model the time series of fMRI signals, allowing for the estimation of GLM coefficients using maximum likelihood estimation and alternating optimization, which accounts for variable conditional variance, and applies hypothesis testing using a T-test to determine brain activity.

Benefits of technology

This approach effectively mitigates the impact of heteroscedasticity, improving the precision of brain activity identification by providing more accurate estimates and increasing the number of correctly identified active voxels, while maintaining similar sensitivity to conventional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure IB2023054815_14112024_PF_FP_ABST
    Figure IB2023054815_14112024_PF_FP_ABST
Patent Text Reader

Abstract

A method for activity identification of a brain region. The method includes obtaining a functional magnetic resonance imaging (fMRI) signal, estimating a plurality of coefficients for a general linear model (GLM) of the fMRI signal, and identifying an activity of the brain region by applying a hypothesis testing to the plurality of coefficients. The fMRI signal is obtained utilizing an fMRI device. Obtaining the fMRI signal includes scanning a brain of a subject performing a task. The fMRI signal includes a time series associated with the brain region. Estimating the plurality of coefficients includes maximizing a likelihood function of the fMRI signal. The likelihood function includes a likelihood of the time series under a generalized autoregressive conditionally heteroscedastic (GARCH) process. Identifying the activity includes applying a hypothesis testing to the plurality of coefficients.
Need to check novelty before this filing date? Find Prior Art

Description

333-8314 ACTIVITY IDENTIFICATION OF A BRAIN REGION TECHNICAL FIELD

[0001] The present disclosure generally relates to medical signal processing, and particularly, to brain activity identification by processing functional magnetic resonance imaging signals. BACKGROUND ART

[0002] In recent years, imaging techniques have enabled data collection on human brain function. In functional magnetic resonance imaging (fMRI), blood oxygen level in different points of brain is measured, and a three-dimensional image of relative blood oxygen level– dependent (BOLD) is generated at each moment. fMRI is based on a fact that changes in BOLD signal may indicate changes in neural cell activity, as oxygen level in blood vessels adjacent to active neural cells decreases. In fMRI, a spatial resolution for observing oxygen status is less than two millimeters, and a temporal resolution is more than a few seconds.

[0003] Despite the widespread applications of fMRI, fMRI signal processing is subject to significant challenges. fMRI data may consist sequences of 100 to 500 three-dimensional images, each with more than 10,000 points (voxels). Different data points of fMRI signals are of high spatiotemporal correlation. Moreover, noise in fMRI signal is very high, such that a target pattern (to be identified in fMRI signal) covers less than 2 to 3 percent of total changes in fMRI time series. Besides, different sources of noise may impact fMRI signal, including noise from imaging device, head motion during imaging, respiratory pattern, heartbeat pattern, changes in blood flow rate, and unpredictable cognitive brain activities.

[0004] Identification of brain regions that become active in response to external stimuli are referred to as active stimulation points, and is crucial in fMRI-based neuroimaging [US patent No. 8861815 B2]. Neuroscientists and physicians use this tool to identify locations of activation in vital brain activities such as thinking, speech, movement, emotion, and attention, to determine side effects of brain stroke, cancer, and neurodegenerative diseases (such as Alzheimer's) on brain function, to monitor a progress of a disease and treatment, and to have a more precise presurgical planning.

[0005] Correlation method is conventional for identification and evaluation of active points in fMRI data [US patent No. 9072496B B2]. Correlation method is based on general linear model (GLM). GLM is capable of dealing with a small number of fMRI images and is of a333-8314 high statistical interpretability. Besides, uncertainty levels are known in each step of derivations in GLM. Therefore, GLM-based correlation method is most prominent tool in fMRI signal processing. In correlation method, after preprocessing of fMRI signals, brain images are examined voxel by voxel to accurately model and describe ideal signal in an active voxel using GLM. Two main reasons for univariate analyses of fMRI data include: first, a number of voxels is much larger than a number of images, making analysis of spatial relationships challenging; and second, specific research hypotheses are often posed for each region or even a single voxel instead of analyzing entire brain, making inference and reporting difficult for just one point in multivariate analyses. Subsequently, a similarity between observed image (at a desired point) and an ideal signal is measured by applying statistical tests such that voxels with similar values are identified as voxels related to brain activity. Finally, different results of active and inactive voxels in a two-dimensional image are placed side by side, and a three-dimensional image of active brain regions is generated.

[0006] Noise distribution impacts statistical testing of fMRI signals. Many imaging conditions in fMRI that are unknown and unmeasurable are considered as noise factors and may be approximated by assuming specific probability distributions. A common assumption in GLM approach and related statistical tests is homoscedasticity, that is, a conditional variance (current sample variance given past samples in a time series) of noise is constant. However, recent evidences reveal that homoscedasticity may not be valid under certain conditions, indicating that the conditional variance of fMRI signals may change over time, thereby affecting statistical testing of fMRI signals. Changes in conditional variance of fMRI signals may arise from multiple sources. First factor is rapid head motion, leading to a phenomenon known as spin history. As a result of rapid head motion, excited proton spins in different regions of brain may change, resulting in changing statistical properties of fMRI signals, particularly variance. Second factor is due to slow head movements, leaving some residual effects on fMRI signals and affecting variance, particularly in areas where brain tissue changes. Third factor is scanner noise, that is, effects of the fMRI device. Statistically, if conditional variance of fMRI signals change, values of remaining GLM parameters are affected, and a distribution of GLM residuals may change from a simple (e.g., Gaussian) distribution to a complex, heavy-tailed distribution, leading to biased parameter estimates and incorrect statistical inference in fMRI studies.333-8314

[0007] Various methods are developed to address variable conditional variance. Some methods are only applicable to cases that variance variability is similar in all brain regions. In practice, however, noise variance may change in one region of brain but not in other regions. Therefore, assuming similar variance variations in all brain regions may limit a precision of brain activity identification. Other methods may be based on a Bergomi’s stochastic volatility model. The Bergomi’s stochastic volatility model may require a large number of fMRI images for a precise activity identification. However, a number of fMRI images may be limited to 100 to 500, making parameter estimation challenging.

[0008] There is, therefore, a need for a GLM-based brain activity identification method that copes with variance variability in different brain regions and different fMRI images, and also is applicable to a small number of fMRI images. SUMMARY OF THE DISCLOSURE

[0009] This summary is intended to provide an overview of the subject matter of the present disclosure, and is not intended to identify essential elements or key elements of the subject matter, nor is it intended to be used to determine the scope of the claimed implementations. The proper scope of the present disclosure may be ascertained from the claims set forth below in view of the detailed description below and the drawings.

[0010] In one general aspect, the present disclosure describes an exemplary method for activity identification of a brain region. An exemplary method may include obtaining a functional magnetic resonance imaging (fMRI) signal, estimating a plurality of coefficients for a general linear model (GLM) of the fMRI signal, and identifying an activity of the brain region by applying a hypothesis testing to the plurality of coefficients. An exemplary fMRI signal may be obtained utilizing an fMRI device. Obtaining an exemplary fMRI signal may include scanning a brain of a subject performing a task. An exemplary fMRI signal may include a time series associated with the brain region. In an exemplary embodiment, the plurality of coefficients may be estimated utilizing one or more processors. In an exemplary embodiment, estimating the plurality of coefficients may include maximizing a likelihood function of the fMRI signal. An exemplary likelihood function may include a likelihood of the time series under a generalized autoregressive conditionally heteroscedastic (GARCH) process. An exemplary activity of the brain region may be identified utilizing the one or more333-8314 processors. Identifying an exemplary activity may include applying a hypothesis testing to the plurality of coefficients.

[0011] Maximizing an exemplary likelihood function may include minimizing a negative log-likelihood function of the likelihood function. Minimizing an exemplary negative log- likelihood function may include initializing a 0^^ℎplurality of GARCH parameters of the GARCH process and repeating an iterative process until a termination condition is satisfied. An ^^^^ℎiteration of an exemplary iterative process where ^^ ≥ 1 may include obtaining an ^^^^ℎplurality of GLM parameters given an ( ^^ − 1)^^ℎplurality of GARCH parameters and obtaining an ^^^^ℎplurality of GARCH parameters given the ^^^^ℎplurality of GLM parameters. An exemplary ^^^^ℎplurality of GLM parameters may include the plurality of coefficients, where ^^ is a number of iterations for the iterative process to be terminated.

[0012] In an exemplary embodiment, obtaining the ^^^^ℎplurality of GLM parameters may include computing a plurality of variances of the time series and computing the ^^^^ℎplurality of GLM parameters based on the plurality of variances. In an exemplary embodiment, the plurality of variances may be computed from the ( ^^ − 1)^^ℎplurality of GARCH parameters. In an exemplary embodiment, each of the plurality of variances may be associated with a variance of the time series at a respective time instant.

[0013] In an exemplary embodiment, obtaining the ^^^^ℎplurality of GARCH parameters may include minimizing an objective function with respect to the ^^^^ℎplurality of GARCH parameters subject to a set of constraints. Applying an exemplary hypothesis testing to the plurality of coefficients may include applying a T-test to the plurality of coefficients.

[0014] Other exemplary systems, methods, features, and advantages of the implementations will be, or will become, apparent to one of ordinary skill in the art upon examination of the following figures and detailed description. It is intended that all such additional systems, methods, features and advantages be included within this description and this summary, be within the scope of the implementations, and be protected by the claims herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawing figures depict one or more implementations in accord with the present teachings, by way of example only, not by way of limitation. In the figures, like reference numerals refer to the same or similar elements.333-8314

[0016] FIG. 1A shows a flowchart of a method for activity identification of a brain region, consistent with one or more exemplary embodiments of the present disclosure.

[0017] FIG. 1B shows a flowchart of a method for estimating a plurality of coefficients, consistent with one or more exemplary embodiments of the present disclosure.

[0018] FIG. 1C shows a flowchart of a method for repeating an iterative process, consistent with one or more exemplary embodiments of the present disclosure.

[0019] FIG. 1D shows a flowchart of a method for obtaining a plurality of GLM parameters, consistent with one or more exemplary embodiments of the present disclosure.

[0020] FIG. 2 shows a schematic of a system for activity identification of a brain region, consistent with one or more exemplary embodiments of the present disclosure.

[0021] FIG. 3 shows a high-level functional block diagram of a computer system, consistent with one or more exemplary embodiments of the present disclosure.

[0022] FIG. 4 shows active brain regions for three subjects, consistent with one or more exemplary embodiments of the present disclosure.

[0023] FIG. 5 shows heteroskedasticity behavior of residual errors for fMRI signals of three different subjects, consistent with one or more exemplary embodiments of the present disclosure. DESCRIPTION OF EMBODIMENTS

[0024] In the following detailed description, numerous specific details are set forth by way of examples in order to provide a thorough understanding of the relevant teachings. However, it should be apparent that the present teachings may be practiced without such details. In other instances, well known methods, procedures, components, and / or circuitry have been described at a relatively high-level, without detail, in order to avoid unnecessarily obscuring aspects of the present teachings.

[0025] The following detailed description is presented to enable a person skilled in the art to make and use the methods and devices disclosed in exemplary embodiments of the present disclosure. For purposes of explanation, specific nomenclature is set forth to provide a thorough understanding of the present disclosure. However, it will be apparent to one skilled in the art that these specific details are not required to practice the disclosed exemplary embodiments. Descriptions of specific exemplary embodiments are provided only as representative examples. Various modifications to the exemplary implementations will be333-8314 readily apparent to one skilled in the art, and the general principles defined herein may be applied to other implementations and applications without departing from the scope of the present disclosure. The present disclosure is not intended to be limited to the implementations shown, but is to be accorded the widest possible scope consistent with the principles and features disclosed herein.

[0026] Herein is disclosed an exemplary method for activity identification of a brain region. The method starts with obtaining a functional magnetic resonance imaging (fMRI) signal by scanning a brain of a subject while performing a task. An exemplary fMRI signal may include a times series data of a brain region of the subject. Next, a number of coefficients representing the fMRI signal in a general linear model (GLM) are estimated. Each exemplary GLM coefficient may include a contribution of a respective regressor in the GLM. Exemplary GLM coefficients may be estimated by maximum likelihood estimation (MLE) approach. In other words, exemplary GLM coefficients may be estimated such that a likelihood function of the fMRI signal is maximized. MLE depends on a noise distribution of the fMRI signal. To capture a variable conditional variance that is present in typical fMRI signals, a generalized autoregressive conditionally heteroscedastic (GARCH) process is considered for a time series of the fMRI signal. Since both GLM coefficients and GARCH process parameters are unknown, an alternating optimization approach is employed for likelihood maximization, wherein GLM coefficients are obtained having the GARCH process parameters fixed and vice versa.

[0027] After estimating GLM coefficients, a hypothesis testing is applied to the GLM coefficients, determining activity or inactivity of the brain region based on a null hypothesis. An exemplary null hypothesis may include a linear combination of GLM coefficients being zero. An exemplary hypothesis testing may include a T-test that is evaluated by a T-statistic. An exemplary closed-form expression for the T-statistic is obtained. Finally, according to a resulting p-value and a predetermined significance level, activity or inactivity of the brain region is determined.

[0028] FIG. 1A shows a flowchart of a method for activity identification of a brain region, consistent with one or more exemplary embodiments of the present disclosure. In an exemplary embodiment, a method 100 may include obtaining an fMRI signal (step 102), estimating a plurality of coefficients for a GLM of the fMRI signal (step 104), and identifying an activity of the brain region by applying a hypothesis testing to the plurality of coefficients333-8314 (step 106). In an exemplary embodiment, method 100 may identify activity or inactivity of a brain region in response to certain stimuli. By applying method 100 to different brain regions of a brain of a subject, method 100 may provide an exemplary three-dimensional image of the brain that illustrates active and inactive parts of the brain in response to stimuli.

[0029] FIG. 2 shows a schematic of a system for activity identification of a brain region, consistent with one or more exemplary embodiments of the present disclosure. In an exemplary embodiment, a system 200 may include an fMRI device 202 and a processor 204. Referring to FIGs. 1A and 2, in an exemplary embodiment, different steps of method 100 may be implemented utilizing system 200.

[0030] For further detail with respect to step 102, in an exemplary embodiment, the fMRI signal may be obtained utilizing fMRI device 202. In an exemplary embodiment, fMRI device 202 may include a radio frequency (RF) coil 206, a gradient coil 208, and a magnet 210. An exemplary brain of a subject 212 may be exposed to fMRI device 202 when subject 212 lies into a scanner chamber of fMRI device 202. In an exemplary embodiment, fMRI device 202 may utilize a magnetic field and radio waves to generate an fMRI signal. When subject 212 enters magnet 210, an exemplary magnetic field may align protons in water molecules in a body of subject 212, causing water molecules to emit an RF signal. In an exemplary embodiment, gradient coil 208 may be used to generate a magnetic field that varies across space. Gradient coil 208 may allow fMRI device 202 to generate images of specific areas of the subject’s brain. In an exemplary embodiment, RF coil 206 may detect the RF signal and may use the RF signal to generate the fMRI signal. An exemplary RF signal may then be digitized and sent to processor 204 for further processing. An exemplary fMRI signal may include a time series of digitized RF signal for a brain region of subject 212. In an exemplary embodiment, processor 204 may control RF coil 206 through RF modules 214 including an RF pulse generator, a digitizer, an RF source, and an amplifier. In an exemplary embodiment, processor 204 may control gradient coil 206 through gradient amplifiers 216.

[0031] An exemplary fMRI signal may include a blood oxygen level-dependent (BOLD) contrast signal. An exemplary BOLD contrast signal may represent magnetic properties of oxygenated and deoxygenated hemoglobin in subject’s blood. Therefore, an exemplary BOLD signal may be used to identify active brain regions of subject 212 because active neurons consume more oxygen than inactive neurons.333-8314

[0032] Obtaining an exemplary fMRI signal may include scanning a brain of subject 212 performing a task. During an fMRI scan, subject 212 may be asked to perform a task, such as viewing pictures or solving problems. As the brain becomes more active, blood flow to a brain region corresponding to the task may increase, delivering more oxygen and glucose to neurons of the brain region. As a result, increase in blood flow may cause a change in magnetic properties of the blood flow and may be detected by fMRI device 202.

[0033] Obtain an exemplary fMRI signal may further include a number of preprocessing steps on the digitized RF signal. Exemplary preprocessing steps may be performed utilizing processor 204. Preprocessing may be referred to as a set of data processing steps applied to fMRI data before statistical analysis. Exemplary preprocessing steps may minimize an impact of noise and artifacts on fMRI signal analysis and to improve a detection performance of functional activity in the brain. Exemplary preprocessing steps in fMRI may include slice timing correction, motion correction, spatial normalization, smoothing, and high-pass filtering. An exemplary slice timing correction may adjust for time differences in image acquisition between slices in a voxel to account for potential effect of slice order on results. An exemplary motion correction may correct for a subject's head motion during a scan. A correction may involve realigning all voxels to a reference voxel to minimize an effect of motion on results. An exemplary spatial normalization may map functional data to a standard anatomical space, helping to combine data from different subjects and may study and improve an alignment of functional data with corresponding anatomical structures. An exemplary smoothing may apply a spatial filter to fMRI data, reducing an amount of noise in the fMRI data. An exemplary smoothing kernel size may be selected based on an expected size of brain activation regions. An exemplary high-pass filtering may remove low-frequency noise from the fMRI data and may improve a sensitivity of an analysis to a task-related signal.

[0034] In further detail regarding step 104, in an exemplary embodiment, the plurality of coefficients may be estimated utilizing processor 204. An exemplary fMRI signal may include a time series ^^ of a single voxel at ^^ different time instants, that is, ^^ =[^^(1), ⋯ ^^(^^)] ^^. An exemplary design matrix ^^ may consist a set of regressors as columns of the design matrix that represent different experimental conditions or stimuli. Each exemplary regressor may correspond to a time course of a particular condition or stimulus, and values in a respective column may represent a timing and duration of that condition or333-8314 stimulus. In an exemplary embodiment, design matrix ^^ may include two sets of regressors, one corresponding to stimulation functions ^^^^and another corresponding to drift patterns ^^^^of fMRI device 202. In an exemplary embodiment, stimulation functions ^^^^may describe an experimental design of an fMRI signal acquisition. Exemplary stimulation functions may be used to model a relationship between a neural activity of the brain region and the fMRI signal. Besides, in an exemplary embodiment, drift patterns ^^^^, also called nuisance regressors, may represent sources of noise in the fMRI signal. Exemplary drift patterns may include motion parameters, physiological signals, and other sources of noise that affect the fMRI signal. An exemplary fMRI signal of a voxel in a GLM may be defined according to the following: ^^ = + ^^ Equation (1)where ^^ =[^^^^^^^^] ^^is a vector of the plurality of coefficients, ^^^^is a subset of coefficients corresponding to stimulation functions ^^^^, ^^^^is a subset of coefficients corresponding to drift patterns ^^^^, and ^^ is a vector of noise values at ^^ different time instants. In an exemplary embodiment, to capture conditional variance variability in time series ^^, noise vector ^^ is assumed to be Gaussian-like distributed with zero mean and variable variance at different time instants. In other words, in an exemplary embodiment, noise vector ^^ is of a GARCH process. Therefore, in an exemplary embodiment, noise vector ^^ may be defined according to the following: Equation (2)where: ^^( ^^) is a noise value in time series ^^ a time instant ^^, ℎ( ^^) is a standard deviation of noise at time instant ^^, and ^^( ^^) is a zero-mean white Gaussian noise at time instant ^^.

[0035] In an exemplary embodiment, a variance of noise at time instant ^^ may depend on variances and noise values at previous time instants. In an exemplary embodiment, a variance of noise at time instant ^^ may be defined according to the following:Equation (3)333-8314 where ^^ is a number of lag variances of the GARCH process and ^^ is a number of lag residual errors of the GLM. An exemplary GARCH process may be defined by a plurality of GARCH parameters ^^ =⋯ , ^^^^, ^^1, ⋯ , ^^^^]. By defining a vector of previous noise and variance values ^^(^^)= [1, ^^2(^^ − 1), ⋯ , ^^2(^^ − ^^), ℎ2(^^ − 1), ⋯ , ℎ2( ^^ − ^^)], an exemplary variance at time instant ^^ may be given as ℎ2( ^^) = ^^( ^^) ^^.

[0036] In an exemplary embodiment, estimating plurality of coefficients ^^ may include maximizing a likelihood function of the fMRI signal. An exemplary likelihood function may include a likelihood of time series ^^ under the GARCH process defined in Eq. (2) and Eq. (3). An exemplary likelihood function may be defined according to the following: ^^( ^^, ^^; ^^)Equation (4)

[0037] In an exemplary embodiment, maximizing a likelihood function ^^ may include minimizing a negative log-likelihood function of likelihood function ^^. According to Eq. (4) and by some approximations, an exemplary negative log-likelihood function may be defined according to the following:Equation (5) where ^^( ^^) ∈ ^^^^×1is a ^^^^ℎrow of design matrix ^^ and ^^ is a number of regressors in design matrix ^^. In an exemplary embodiment, minimizing negative log-likelihood function ^^ may include jointly optimizing plurality of coefficients ^^ and plurality of GARCH parameters ^^. Instead of a joint optimization, an exemplary alternating optimization may be employed in a practical implementation, as described below.

[0038] FIG. 1B shows a flowchart of a method for estimating a plurality of coefficients, consistent with one or more exemplary embodiments of the present disclosure. Minimizing an exemplary negative log-likelihood function may include initializing a 0^^ℎplurality of GARCH parameters of the GARCH process (step 108) and repeating an iterative process until a termination condition is satisfied (step 110).

[0039] For further detail with regard to step 108, in an exemplary embodiment, an alternating optimization of plurality of coefficients ^^ and plurality of GARCH parameters ^^ may require initialization of plurality of GARCH parameters ^^. Therefore, in an exemplary embodiment, plurality of GARCH parameters ^^ may be initialized by the 0^^ℎplurality of GARCH parameters. An exemplary 0^^ℎplurality of GARCH parameters may include a number of333-8314 positive random values such that a summation of the 0^^ℎplurality of GARCH parameters is equal to one.

[0040] In further detail with respect to step 110, FIG. 1C shows a flowchart of a method for repeating an iterative process, consistent with one or more exemplary embodiments of the present disclosure. An ^^^^ℎiteration of an exemplary iterative process where ^^ ≥ 1 may include obtaining an ^^^^ℎplurality of GLM parameters given an ( ^^ − 1)^^ℎplurality of GARCH parameters (step 112) and obtaining an ^^^^ℎplurality of GARCH parameters given the ^^^^ℎplurality of GLM parameters (step 114). An exemplary iterative process may be employed for alternating optimization of plurality of coefficients ^^ and plurality of GARCH parameters ^^.

[0041] For further detail regarding step 112, FIG. 1D shows a flowchart of a method for obtaining a plurality of GLM parameters, consistent with one or more exemplary embodiments of the present disclosure. In an exemplary embodiment, obtaining the ^^^^ℎplurality of GLM parameters may include computing a plurality of variances of time series ^^ (step 116) and computing the ^^^^ℎplurality of GLM parameters based on the plurality of variances (step 118). In an exemplary ^^^^ℎiteration of the iterative process, an ^^^^ℎplurality of GLM parameters ^^( ^^)may be obtained by minimizing function ^^ defined in Eq. (5). In doing so, in an exemplary embodiment, ^^^^ℎplurality of GLM parameters ^^( ^^)may be obtained by setting a derivative of function ^^ with respect to ^^( ^^)to zero. In other words, ^^^^ℎplurality of GLM parameters ^^( ^^)may be obtained by solving an exemplary equation defined according to the following:Equation (6)

[0042] In further detail with regard to step 116, in an exemplary embodiment, solving Eq. (6) with respect to ^^ may require values of the plurality of variances, that is, ℎ2(1), ⋯ ,Therefore, exemplary values of ℎ2( ^^) may be successively calculated according to Eq. (3) given an ( ^^ − 1)^^ℎplurality of GARCH parametersand a positive random value or a zero value for ℎ2( ^^ = 0).

[0043] For further detail with respect to step 118, in an exemplary embodiment, ^^^^ℎplurality of GLM parameters ^^( ^^)may be computed by solving Eq. (6) and according to an operation defined by the following:Equation (7)333-8314is a diagonal matrix with diagonal elements including inverses of the plurality of variances, that is, ^^( ^^−1)= diag([ℎ(1)−2, ℎ(2)−2, … , ℎ( ^^)−2]) ∈ ^^^^× ^^and diag( ^^) is a diagonal matrix with diagonal elements equal to a vector ^^. In an exemplary embodiment, a ^^^^ℎplurality of GLM parameters ^^( ^^)may be equal to plurality of coefficients ^^, where ^^ is a number of iterations for the iterative process to be terminated.

[0044] Referring to FIG. 1C, in an exemplary embodiment, step 114 may include obtaining an ^^^^ℎplurality of GARCH parameters ^^( ^^)given ^^^^ℎplurality of GLM parameters ^^( ^^). In an exemplary embodiment, obtaining ^^^^ℎplurality of GARCH parameters ^^( ^^)may include minimizing an objective function with respect to ^^^^ℎplurality of GARCH parameters ^^( ^^)subject to a set of constraints. By plugging ℎ2( ^^) = ^^( ^^) ^^( ^^)into Eq. (5), an exemplary objective function ^^ ^^ may be defined according to the following: 2 ^^^2log( ^^( ^^)^̂(^^)(^^)) + ( ^^( ^^)2^^ ) ] Equation (8)( ^^)whereis a residual error of the GLM at a time instant ^^. An exemplary set of constraints may include a first constraint and a second constraint. An exemplary first constraint may be defined according to an inequality − ^^^^^^ < ^^ , where ^^^^is an identity matrix and ^^ = ^^ + ^^ + 1. An exemplary second constraint may be defined according to an inequality < 1 , where ^^ = [0,1,1, … ,1] is a row vector of length ^^ . In an exemplary embodiment, objective function ^^ ^^ may be efficiently minimized subject to the set of constraints utilizing interior point methods.

[0045] Referring to FIGs. 1A and 2, in an exemplary embodiment, step 106 may include applying a hypothesis testing to plurality of coefficients ^^. An exemplary hypothesis testing may be applied to plurality of coefficients ^^ utilizing processor 204. Applying an exemplary hypothesis testing may require statistics of the plurality of coefficients. In an exemplary embodiment, a distribution of the plurality of coefficients may be obtained according to the following: ^^̂~ ^^( ^^, ^^2(( ^^ ^^)^^^^ ^^)− ^^) = ^^( ^^, ^^2( ^^^^^^^^^^ ^^)− ^^) Equation (9) where ^^( ^^, ^^) is a normal distribution with mean ^^ and covariance matrix ^^, ^^ may be obtained such that ^^ ^^^^= ^^, and ^^2may be obtained according to an operation defined by the following:Equation (10)333-8314

[0046] Applying an exemplary hypothesis testing to the plurality of coefficients may include applying a T-test to plurality of coefficients ^^. An exemplary hypothesis testing may include a null hypothesis defined according to ^^^^^^ = 0. In an exemplary embodiment, a vector ^^ may be selected to determine whether a certain coefficient of plurality of coefficients ^^ is larger than zero or to determine a coefficient that is significantly larger than other coefficients in statistical sense. Applying an exemplary T-test to plurality of coefficients ^^ may include computing a T-statistic as a function of plurality of coefficients ^^ and vector ^^ . An exemplary T-statistics may be defined according to an operation defined by the following Equation (11)

[0047] In an exemplary embodiment, having a distribution of ^^̂ as in Eq. (9), a value of the T-statistics may be obtained according to the following: ^^^^ ^^ ^^ ^^ ^^−Equation (12)

[0048] In an exemplary embodiment, after computation of ^^^^ ^^ ^^ ^^ ^^− ^^ ^^ ^^from Eq. (12), a p- value may be derived from a t-distribution with ^^ − ^^ degrees of freedom. An exemplary null hypothesis may be rejected, that is, the brain region is decided to be inactive, responsive to the p-value being smaller than a significance level. In contrast, an exemplary null hypothesis may be accepted, that is, the brain region is decided to be active, responsive to the p-value being larger than the significance level. An exemplary significance level may be set equal to 0.05.

[0049] FIG. 3 shows an example computer system 300 in which an embodiment of the present invention, or portions thereof, may be implemented as computer-readable code, consistent with exemplary embodiments of the present disclosure. For example, different steps of method 100 may be implemented in computer system 300 using hardware, software, firmware, tangible computer readable media having instructions stored thereon, or a combination thereof and may be implemented in one or more computer systems or other processing systems. Hardware, software, or any combination of such may embody any of the modules and components in FIGs.1A-2.

[0050] If programmable logic is used, such logic may execute on a commercially available processing platform or a special purpose device. One ordinary skill in the art may appreciate that an embodiment of the disclosed subject matter can be practiced with various computer333-8314 system configurations, including multi-core multiprocessor systems, minicomputers, mainframe computers, computers linked or clustered with distributed functions, as well as pervasive or miniature computers that may be embedded into virtually any device.

[0051] For instance, a computing device having at least one processor device and a memory may be used to implement the above-described embodiments. A processor device may be a single processor, a plurality of processors, or combinations thereof. Processor devices may have one or more processor “cores.”

[0052] An embodiment of the invention is described in terms of this example computer system 300. After reading this description, it will become apparent to a person skilled in the relevant art how to implement the invention using other computer systems and / or computer architectures. Although operations may be described as a sequential process, some of the operations may in fact be performed in parallel, concurrently, and / or in a distributed environment, and with program code stored locally or remotely for access by single or multi- processor machines. In addition, in some embodiments the order of operations may be rearranged without departing from the spirit of the disclosed subject matter.

[0053] Processor device 304 may be a special purpose (e.g., a graphical processing unit) or a general-purpose processor device. As will be appreciated by persons skilled in the relevant art, processor device 304 may also be a single processor in a multi-core / multiprocessor system, such system operating alone, or in a cluster of computing devices operating in a cluster or server farm. Processor device 304 may be connected to a communication infrastructure 306, for example, a bus, message queue, network, or multi-core message- passing scheme.

[0054] In an exemplary embodiment, computer system 300 may include a display interface 302, for example a video connector, to transfer data to a display unit 330, for example, a monitor. Computer system 300 may also include a main memory 308, for example, random access memory (RAM), and may also include a secondary memory 310. Secondary memory 310 may include, for example, a hard disk drive 312, and a removable storage drive 314. Removable storage drive 314 may include a floppy disk drive, a magnetic tape drive, an optical disk drive, a flash memory, or the like. Removable storage drive 314 may read from and / or write to a removable storage unit 318 in a well-known manner. Removable storage unit 318 may include a floppy disk, a magnetic tape, an optical disk, etc., which may be read by and written to by removable storage drive 314. As will be appreciated by persons skilled333-8314 in the relevant art, removable storage unit 318 may include a computer usable storage medium having stored therein computer software and / or data.

[0055] In alternative implementations, secondary memory 310 may include other similar means for allowing computer programs or other instructions to be loaded into computer system 300. Such means may include, for example, a removable storage unit 322 and an interface 320. Examples of such means may include a program cartridge and cartridge interface (such as that found in video game devices), a removable memory chip (such as an EPROM, or PROM) and associated socket, and other removable storage units 322 and interfaces 320 which allow software and data to be transferred from removable storage unit 322 to computer system 300.

[0056] Computer system 300 may also include a communications interface 324. Communications interface 324 allows software and data to be transferred between computer system 300 and external devices. Communications interface 324 may include a modem, a network interface (such as an Ethernet card), a communications port, a PCMCIA slot and card, or the like. Software and data transferred via communications interface 324 may be in the form of signals, which may be electronic, electromagnetic, optical, or other signals capable of being received by communications interface 324. These signals may be provided to communications interface 324 via a communications path 326. Communications path 326 carries signals and may be implemented using wire or cable, fiber optics, a phone line, a cellular phone link, an RF link or other communications channels.

[0057] In this document, the terms “computer program medium” and “computer usable medium” are used to generally refer to media such as removable storage unit 318, removable storage unit 322, and a hard disk installed in hard disk drive 312. Computer program medium and computer usable medium may also refer to memories, such as main memory 308 and secondary memory 310, which may be memory semiconductors (e.g. DRAMs, etc.).

[0058] Computer programs (also called computer control logic) are stored in main memory 308 and / or secondary memory 310. Computer programs may also be received via communications interface 324. Such computer programs, when executed, enable computer system 300 to implement different embodiments of the present disclosure as discussed herein. In particular, the computer programs, when executed, enable processor device 304 to implement the processes of the present disclosure, such as the operations in method 100 illustrated by flowcharts of FIGs. 1A-1D discussed above. Accordingly, such computer333-8314 programs represent controllers of computer system 300. Where an exemplary embodiment of method 100 is implemented using software, the software may be stored in a computer program product and loaded into computer system 300 using removable storage drive 314, interface 320, and hard disk drive 312, or communications interface 324.

[0059] Embodiments of the present disclosure also may be directed to computer program products including software stored on any computer useable medium. Such software, when executed in one or more data processing device, causes a data processing device to operate as described herein. An embodiment of the present disclosure may employ any computer useable or readable medium. Examples of computer useable mediums include, but are not limited to, primary storage devices (e.g., any type of random access memory), secondary storage devices (e.g., hard drives, floppy disks, CD ROMS, ZIP disks, tapes, magnetic storage devices, and optical storage devices, MEMS, nanotechnological storage device, etc.).

[0060] The embodiments have been described above with the aid of functional building blocks illustrating the implementation of specified functions and relationships thereof. The boundaries of these functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternate boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. EXAMPLE

[0061] In this example, performance of a method for activity identification of brain regions (similar to method 100) is demonstrated. fMRI signals are extracted for 10 different subjects while performing an auditory oddball task. Each subject has to identify an odd sound from background sounds. Three different stimuli are randomly played for each subject, including a standard stimulus, a target stimulus to be identified, and a rare stimulus that is a random digital noise. Stimuli are generated in each repetition time with probabilities about 0.6 , 0.075 , and 0.075 , respectively. fMRI signals are simulated based on principles of independent component analysis (ICA), and a GARCH noise with parameters about (0.1, 0.8) is added to simulated fMRI signals. For each subject, 200 images (one image per two seconds) are generated and each image includes 30 × 30 voxels. A contrast to noise of signals is set to about 0.75. A ratio of standard deviation of fMRI signals to standard deviation of noise is set in an interval (0.2, 2.52). Table I compares a power and a sensitivity of the method (called GARCH-GLM) and a conventional method that is GLM-based but333-8314 estimates GLM coefficients for a likelihood function of a noise with constant conditional variance.According to Table I, sensitivity of the GARCH-GLM method is similar to that of the conventional method. However, a power of the GARCH-GLM method is about 14% higher than the conventional method, meaning that a number of correctly identified voxels by the GARCH-GLM method is larger than that of the conventional method while an identification rate of both methods is relatively the same.

[0062] FIG. 4 shows active brain regions for three subjects, consistent with one or more exemplary embodiments of the present disclosure. A graphical representation of identified brain regions for three subjects is shown in FIG.4. First row shows correct map of active regions, second row shows identified active regions by the conventional method, third row shows identified active regions by the GARCH-GLM method, and fourth row shows a difference between identified regions of the GARCH-GLM method and the conventional method. As FIG. 4 shows, the GARCH-GLM method in the third row identifies more active regions than the conventional method.

[0063] FIG. 5 shows heteroskedasticity behavior of residual errors for fMRI signals of three different subjects, consistent with one or more exemplary embodiments of the present disclosure. Heteroscedasticity of residual errors is an efficiency measure of activity333-8314 identification methods. A p-value of each voxel represents a probability of a null hypothesis, that is, absence of variable conditional variance, in residual errors. Frequency of different p- values of the GARCH-GLM method and the conventional method for three different subjects is shown in FIG. 5. In more than 90% of voxels, p-values of the conventional method are smaller than 0.05, that is, a probability of a constant conditional variance is almost zero. Therefore, variable conditional variance is remained in residual errors for the conventional method. In contrast, most of p-values are larger than 0.05 for the GARCH-GLM method. Therefore, there is no sufficient evidence for presence of variable conditional variance in residual errors of the GARCH-GLM method. In other words, the GARCH-GLM method effectively mitigates heteroscedasticity impact on fMRI signals.

[0064] While the foregoing has described what may be considered to be the best mode and / or other examples, it is understood that various modifications may be made therein and that the subject matter disclosed herein may be implemented in various forms and examples, and that the teachings may be applied in numerous applications, only some of which have been described herein. It is intended by the following claims to claim any and all applications, modifications and variations that fall within the true scope of the present teachings.

[0065] Unless otherwise stated, all measurements, values, ratings, positions, magnitudes, sizes, and other specifications that are set forth in this specification, including in the claims that follow, are approximate, not exact. They are intended to have a reasonable range that is consistent with the functions to which they relate and with what is customary in the art to which they pertain.

[0066] The scope of protection is limited solely by the claims that now follow. That scope is intended and should be interpreted to be as broad as is consistent with the ordinary meaning of the language that is used in the claims when interpreted in light of this specification and the prosecution history that follows and to encompass all structural and functional equivalents. Notwithstanding, none of the claims are intended to embrace subject matter that fails to satisfy the requirement of Sections 101, 102, or 103 of the Patent Act, nor should they be interpreted in such a way. Any unintended embracement of such subject matter is hereby disclaimed.

[0067] Except as stated immediately above, nothing that has been stated or illustrated is intended or should be interpreted to cause a dedication of any component, step, feature,333-8314 object, benefit, advantage, or equivalent to the public, regardless of whether it is or is not recited in the claims.

[0068] It will be understood that the terms and expressions used herein have the ordinary meaning as is accorded to such terms and expressions with respect to their corresponding respective areas of inquiry and study except where specific meanings have otherwise been set forth herein. Relational terms such as first and second and the like may be used solely to distinguish one entity or action from another without necessarily requiring or implying any actual such relationship or order between such entities or actions. The terms “comprises,” “comprising,” or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by “a” or “an” does not, without further constraints, preclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0069] The Abstract of the Disclosure is provided to allow the reader to quickly ascertain the nature of the technical disclosure. It is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims. In addition, in the foregoing Detailed Description, it can be seen that various features are grouped together in various implementations. This is for purposes of streamlining the disclosure, and is not to be interpreted as reflecting an intention that the claimed implementations require more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive subject matter lies in less than all features of a single disclosed implementation. Thus, the following claims are hereby incorporated into the Detailed Description, with each claim standing on its own as a separately claimed subject matter.

[0070] While various implementations have been described, the description is intended to be exemplary, rather than limiting and it will be apparent to those of ordinary skill in the art that many more implementations and implementations are possible that are within the scope of the implementations. Although many possible combinations of features are shown in the accompanying figures and discussed in this detailed description, many other combinations of the disclosed features are possible. Any feature of any implementation may be used in combination with or substituted for any other feature or element in any other implementation unless specifically restricted. Therefore, it will be understood that any of the features shown333-8314 and / or discussed in the present disclosure may be implemented together in any suitable combination. Accordingly, the implementations are not to be restricted except in light of the attached claims and their equivalents. Also, various modifications and changes may be made within the scope of the attached claims.

Claims

What is claimed is:

1. A method for activity identification of a brain region, the method comprising: obtaining, utilizing a functional magnetic resonance imaging (fMRI) device, an fMRI signal by scanning a brain of a subject performing a task, the fMRI signal comprising a time series of the brain region; estimating, utilizing one or more processors, a plurality of general linear model (GLM) parameters of the fMRI signal by minimizing a negative log-likelihood function of a likelihood function of the fMRI signal, the likelihood function comprising a likelihood of the time series under a generalized autoregressive conditionally heteroscedastic (GARCH) process, minimizing the negative log-likelihood function comprising: initializing a 0^^ℎplurality of GARCH parameters of the GARCH process; and repeating an iterative process until a termination condition is satisfied, an ^^^^ℎiteration of the iterative process where ^^ ≥ 1 comprising: obtaining an ^^^^ℎplurality of GLM parameters given an(^^ − 1) ^^ℎplurality of GARCH parameters by: computing a plurality of variances of the time series from the ( ^^ − 1) ^^ℎplurality of GARCH parameters, each of the plurality of variances associated with a variance of the time series at a respective time instant; and computing the ^^^^ℎplurality of GLM parameters according to an operation defined by the following:where:vector of the ^^^^ℎplurality of GLM parameters, ^^ is a design matrix of the GLM,is a diagonal matrix with diagonal elements comprising inverses of the plurality of variances, and ^^ is a vector of elements in the time series;wherein a ^^^^ℎplurality of GLM parameters comprises the plurality of coefficients, where ^^ is a number of iterations for the iterative process to be terminated; and obtaining an ^^^^ℎplurality of GARCH parameters given the ^^^^ℎplurality of GLM parameters by minimizing an objective function with respect to the ^^^^ℎplurality of GARCH parameters subject to a set of constraints, wherein: the objective function is defined according to the following: ^^ 1 ∑[log 2(^^=2where: ^^ is a length of the time series, ^^( ^^) = [1, ^^2̂( ^^ − 1), … , ^^2̂( ^^ − ^^), ℎ2( ^^ − 1), … , ℎ2(^^ − ^^)], ^^ is a vector of the ^^^^ℎplurality of GARCH parameters,is a residual error of the GLM at a time instant ^^, ℎ2( ^^) is a variance of the plurality of variances at the time instant ^^, ^^ is a number of lag variances of the GARCH process, and ^^ is a number of lag residual errors of the GLM; and the set of constraints comprises: a first constraint defined according to the following: − ^^^^^^ < ^^ where ^^^^is an identity matrix and ^^ = ^^ + ^^ + 1; and a second constraint defined according to the following: ^^ ^^ < 1row vector of length ^^; andidentifying, utilizing the one or more processors, an activity of the brain region by applying a T-test to the plurality of GLM parameters through computing a T-statistic according to an operation defined by the following:where:^^ is a number of regressors in the design matrix, ^^ is a vector defining a null hypothesis ^^^^^^ = 0, and ^^ is a vector of the plurality of coefficients.

2. A method for activity identification of a brain region, the method comprising: obtaining, utilizing an fMRI device, an fMRI signal by scanning a brain of a subject performing a task, the fMRI signal comprising a time series associated with the brain region; estimating, utilizing one or more processors, a plurality of coefficients for a GLM of the fMRI signal by maximizing a likelihood function of the fMRI signal, the likelihood function comprising a likelihood of the time series under a GARCH process; and identifying, utilizing the one or more processors, an activity of the brain region by applying a hypothesis testing to the plurality of coefficients.

3. The method of claim 2, wherein maximizing the likelihood function comprises minimizing a negative log-likelihood function of the likelihood function by: initializing a 0^^ℎplurality of GARCH parameters of the GARCH process; and repeating an iterative process until a termination condition is satisfied, an ^^^^ℎiteration of the iterative process where ^^ ≥ 1 comprising: obtaining an ^^^^ℎplurality of GLM parameters given an ( ^^ − 1)^^ℎplurality of GARCH parameters, a ^^^^ℎplurality of GLM parameters comprising the plurality of coefficients, where ^^ is a number of iterations for the iterative process to be terminated; and obtaining an ^^^^ℎplurality of GARCH parameters given the ^^^^ℎplurality of GLM parameters.

4. The method of claim 3, wherein obtaining the ^^^^ℎplurality of GLM parameters comprises:computing a plurality of variances of the time series from the ( ^^ − 1)^^ℎplurality of GARCH parameters, each of the plurality of variances associated with a variance of the time series at a respective time instant; and computing the ^^^^ℎplurality of GLM parameters according to an operation defined by the following:where:is a vector of the ^^^^ℎplurality of GLM parameters, ^^ is a design matrix of the GLM,is a diagonal matrix with diagonal elements comprising inverses of the plurality of variances, and ^^ is a vector of elements in the time series.

5. The method of claim 3, wherein obtaining the ^^^^ℎplurality of GARCH parameters comprises minimizing an objective function with respect to the ^^^^ℎplurality of GARCH parameters subject to a set of constraints, wherein: the objective function is defined according to the following:where: ^^ is a length of the time series, ^^( ^^) = [1, ^^2( ^^ − 1), … , ^^2( ^^ − ^^), ℎ2( ^^ − 1), … , ℎ2( ^^ − ^^)], ^^ is a vector of the ^^^^ℎplurality of GARCH parameters, ^^( ^^) = | ^^ −is a residual error of the GLM at a time instant ^^, ℎ2( ^^) is a variance of the plurality of variances at the time instant ^^, ^^ is a number of lag variances of the GARCH process, and ^^ is a number of lag residual errors of the GLM; and the set of constraints comprises: a first constraint defined according to the following: − ^^^^^^ < ^^ where ^^^^is an identity matrix and ^^ = ^^ + ^^ + 1; and a second constraint defined according to the following:^^ ^^ < 1 where ^^ = [0,1,1, … ,1] is a row vector of length ^^.

6. The method of claim 5, wherein applying the hypothesis testing to the plurality of coefficients comprises applying a T-test to the plurality of coefficients.

7. The method of claim 6, wherein applying the T-test to the plurality of coefficients comprises computing a T-statistic according to an operation defined by the following:where:^^ is a number of regressors in the design matrix, ^^ is a vector defining a null hypothesis ^^^^^^ = 0, and ^^ is a vector of the plurality of coefficients.

8. A system for activity identification of a brain region, the system comprising: an fMRI device configured to obtain an fMRI signal by scanning a brain of a subject performing a task, the fMRI signal comprising a time series associated with the brain region; a memory having processor-readable instructions stored therein; and a processor configured to access the memory and execute the processor-readable instructions, which, when executed by the processor configures the processor to perform a method, the method comprising: estimating a plurality of coefficients for a GLM of the fMRI signal by maximizing a likelihood function of the fMRI signal, the likelihood function comprising a likelihood of the time series under a GARCH process; and identifying an activity of the brain region by applying a hypothesis testing to the plurality of coefficients.

9. The system of claim 8, wherein maximizing the likelihood function comprises minimizing a negative log-likelihood function of the likelihood function by: initializing a 0^^ℎplurality of GARCH parameters of the GARCH process; andrepeating an iterative process until a termination condition is satisfied, an ^^^^ℎiteration of the iterative process where ^^ ≥ 1 comprising: obtaining an ^^^^ℎplurality of GLM parameters given an ( ^^ − 1)^^ℎplurality of GARCH parameters, a ^^^^ℎplurality of GLM parameters comprising the plurality of coefficients, where ^^ is a number of iterations for the iterative process to be terminated; and obtaining an ^^^^ℎplurality of GARCH parameters given the ^^^^ℎplurality of GLM parameters.

10. The system of claim 9, wherein obtaining the ^^^^ℎplurality of GLM parameters comprises: computing a plurality of variances of the time series from the ( ^^ − 1)^^ℎplurality of GARCH parameters, each of the plurality of variances associated with a variance of the time series at a respective time instant; and computing the ^^^^ℎplurality of GLM parameters according to an operation defined by the following:where: ^^ℎis a vector of the ^^ plurality of GLM parameters, ^^ is a design matrix of the GLM,is a diagonal matrix with diagonal elements comprising inverses of the plurality of variances, and ^^ is a vector of elements in the time series.

11. The system of claim 9, wherein obtaining the ^^^^ℎplurality of GARCH parameters comprises minimizing an objective function with respect to the ^^^^ℎplurality of GARCH parameters subject to a set of constraints, wherein: the objective function is defined according to the following:where: ^^ is a length of the time series,^^ is a vector of the ^^^^ℎplurality of GARCH parameters,error of the GLM at a time instant ^^, ℎ2( ^^) is a variance of the plurality of variances at the time instant ^^, ^^ is a number of lag variances of the GARCH process, and ^^ is a number of lag residual errors of the GLM; and the set of constraints comprises: a first constraint defined according to the following: − ^^^^^^ < ^^ where ^^^^is an identity matrix and ^^ = ^^ + ^^ + 1; and a second constraint defined according to the following: ^^ ^^ < 1 where ^^ = [0,1,1, … ,1] is a row vector of length ^^.

12. The system of claim 11, wherein applying the hypothesis testing to the plurality of coefficients comprises applying a T-test to the plurality of coefficients.

13. The system of claim 12, wherein applying the T-test to the plurality of coefficients comprises computing a T-statistic according to an operation defined by the following:where:^^ is a number of regressors in the design matrix, ^^ is a vector defining a null hypothesis ^^^^^^ = 0, and ^^ is a vector of the plurality of coefficients.