Classification of functional brain images with a spatio-temporal dissimilarity map
Introduction
Functional magnetic resonance imaging (fMRI) is a noninvasive neuroimaging technique used to study functional activity in the living human brain. The task-evoked signal in fMRI is based on the blood oxygen level dependence (BOLD) effect (Ogawa et al., 1990). Neuronal activity is measured indirectly by recording hemodynamic changes (Jueptner and Weiller, 1995). Data collected on each subject are four-dimensional, with three spatial dimensions measured over time.
This paper addresses the problem of classifying subjects into groups using both temporal and spatial information in fMRI data, which has the potential to assist in early disease detection and diagnosis. For example, mild cognitive impairment (MCI) is a condition that predicts the onset of Alzheimer’s disease. It has been shown via FDG-PET that MCI has a distinct functional signature that is a more accurate predictor of the disease than neuropsychological tests (Chetelat et al., 2005). Another application, from cognitive neuroscience research, is confirmation of hypothesized group differences in BOLD response for distinct cognitive states.
Developing methods for classification of subjects based on neuroimaging data poses significant challenges. Large data size and small signal intensity change inhibit accurate classification. Extracting robust features representative of both spatial and temporal aspects of the data plays a crucial role in success of classification.
There have been a number of efforts to classify subjects into groups based on functional data. Some methods proceed from activation maps generated by the general linear model. Kontos et al. (2003) introduced an approach that uses space-filling curves for mapping 3D space into a linear domain. Wang et al. (2004) used dimension reduction techniques on the space-filling curves for discriminative pattern discovery. Liow et al. (2000) applied linear discriminant analysis on the principal components of PET data to classify HIV-1 seropositive patients into AIDS dementia complex (ADC) and non-ADC groups. Ford et al. (2003) used the same approach for discriminating patients and controls based on fMRI data for Alzheimer’s disease, schizophrenia, and mild traumatic brain injury.
Some techniques focus their analysis on regions of interest (ROI). Bogorodzki et al. (2005) developed a method based on differences in regional brain activity. As part of feature selection for preselected ROIs, mean time intensity curves for voxels correlated with the stimulus are computed and modeled using a mixture of time shifted Gaussian functions. New subjects are classified based on derived feature vectors from multiple ROIs. Pokrajac et al. (2005) introduced statistical distance and neural-network-based methods for classification of brain image data contingent on measures of dissimilarity between 3D probability distributions of ROIs.
Mitchell et al. (2004) applied machine learning methods to the classification of cognitive states based on fMRI data, a topic closely related to classification of subjects. They investigated several feature selection and classification approaches. Following the same framework, Zhang et al. (2005) applied machine learning techniques to classification of subjects based on activation maps. In both applications, Gaussian Naive Bayes, Support Vector Machines, and k nearest neighbor classifiers were examined.
While these methods are promising, most rely on contrast maps or predefined regions of interest and are not designed for detecting subtle temporal differences. This paper examines the feasibility of working directly with time series either from whole brain volumes or regions of interest suggested by results from previous studies to select voxels that exhibit highly discriminating features (based on a measure of temporal dissimilarity) and assign group membership to new subjects.
Section snippets
Methods
For this work, functional data are assumed to be motion-corrected, smoothed, and normalized to a standard anatomical template.
We address the problem of classifying subjects by a general methodology that takes two steps. The first is a feature selection step that identifies patterns predictive of group differences by localizing areas in space where temporal behavior is most dissimilar between groups. The second step is classification, where new subjects are assigned group membership based on
Simulated data
Feature selection based on smooth spatio-temporal dissimilarity maps and classification performance were evaluated on simulated data. To ensure a simulation with realistic noise, a task waveform was superimposed on actual fMRI scans of a single subject taken during a rest condition. All computations were performed using Matlab 7.0 (http://www.mathworks.com).
The data were acquired on a 3 T Siemens Allegra scanner in accordance with the Institutional Review Board at the University of Illinois at
Conclusion
This paper presented a unified feature selection and classification procedure for classifying subjects into groups based on four dimensional spatio-temporal data. Unlike previous approaches, the present approach offers the ability to locate spatial regions with temporal differences between groups. The proposed method simultaneously accounts for and identifies intergroup spatial and temporal variability. It uses temporal similarity as a spatially localized measure of similarity, and it
Acknowledgments
We thank the anonymous reviewers for helpful comments on the earlier version of the manuscript.
References (37)
- et al.
Cognition and control in schizophrenia: a computational model of dopamine and prefrontal function
Biol. Psychiatry
(1999) - et al.
Comparing extent of activation: a robust permutation approach
NeuroImage
(2005) - et al.
Thresholding of statistical maps in functional neuroimaging using the false discovery rate
NeuroImage
(2002) - et al.
Validating cluster size inference: random field and permutation methods
NeuroImage
(2003) - et al.
Review: does measurement of regional cerebral blood flow reflect synaptic activity?—Implications for PET and fMRI
NeuroImage
(1995) - et al.
Group analysis in functional neuroimaging: selecting subjects using similarity measures
NeuroImage
(2003) - et al.
The relative involvement of anterior cingulated and prefrontal cortex in attentional control depends on nature of conflict
Cogn. Brain Res.
(2001) - et al.
Applying spatial distribution analysis techniques to classification of 3D medical images
Artif. Intell. Med.
(2005) - et al.
Statistical tests for fMRI based on experimental randomization
NeuroImage
(2003) - et al.
The anterior cingulate as a conflict monitor: fMRI and ERP studies
Physiol. Behav.
(2002)
Selective deficits in prefrontal cortex function in medication-naive patients with schizophrenia
Arch. Gen. Psychiatry
Controlling the false discovery rate: a practical and powerful approach to multiple testing
J. R. Stat. Soc., Ser. B Methodol.
Structural group classification technique based on regional fMRI BOLD responses
IEEE Trans. Med. Imaging
Body-image aberration in schizophrenia
J. Abnorm. Psychol.
FDG-PET measurement is more accurate than neuropsychological assessments to predict global cognitive deterioration in patients with mild cognitive impairment
Neurocase
Context, cortex, and dopamine: a connectionist approach to behavior and biology in schizophrenia
Psychol. Rev.
Paying attention to emotion: an fMRI investigation of cognitive and emotional Stroop tasks
Cogn. Affect. Behav. Neurosci.
Optimal experimental design for event-related fMRI
Hum. Brain Mapp.
Cited by (52)
Generalised spatial and spatiotemporal autoregressive conditional heteroscedasticity
2018, Spatial StatisticsAn adaptive RV measure based fuzzy weighting subspace clustering (ARV-FWSC) for fMRI data analysis
2015, Biomedical Signal Processing and ControlA texture-based method for classification of schizophrenia using fMRI data
2015, Biocybernetics and Biomedical EngineeringCitation Excerpt :The fMRI activation levels have been used in several research efforts to discriminate schizophrenia patients and healthy controls. Identifying groups of voxels and showing between-group temporal dissimilarity has been studied by Shinkareva et al. [12]. They worked directly with fMRI time series from those voxels for classification purposes.
Integration of structural and functional magnetic resonance imaging improves mild cognitive impairment detection
2013, Magnetic Resonance ImagingCitation Excerpt :In this study, the feature vector elements were based on volumetric information in the MTL region (extracted automatically using FreeSurfer), chemical biomarkers in the CSF, and a set of neuropsychological and behavioral measures from various cognitive tests [14]. Abnormal hyper- or hypo-activity within affected brain regions of patients with neuropsychiatric or neurodegenerative conditions has been proposed as a potentially viable fMRI biomarker for predicting illnesses in an individual [19,20]. For example, Tripoliti and colleagues (2010) investigated neuronal activation patterns using fMRI data acquired from healthy young, healthy elderly, and mild AD subjects (n = 41) with a general linear model (GLM) approach that used covariates such as the age of the subject.
Automated classification of fMRI during cognitive control identifies more severely disorganized subjects with schizophrenia
2012, Schizophrenia ResearchCitation Excerpt :The application of pattern classifiers to fMRI data has demonstrated notable sensitivity in detecting subtle abnormalities in the distributed pattern of neural activity in schizophrenia (Yoon et al., 2008a, 2008b). A number of published studies have demonstrated the feasibility of using fMRI data for classifying schizophrenia (Shinkareva et al., 2006; Calhoun et al., 2008; Demirci et al., 2008a; Shen et al., 2010; Costafreda et al., 2011; Costafreda et al.; ). Most have utilized images of unstructured mental states i.e. resting state, (Shinkareva et al., 2006; Calhoun et al., 2008; Shen et al., 2010), and few have utilized images of structured cognition (Demirci et al., 2008a; Costafreda et al., 2011).