Jiacai Zhang

Comparative Study of SVM Methods Combined with Voxel Selection for Object Category Classification on fMRI Data

Background Support vector machine (SVM) has been widely used as accurate and reliable method to decipher brain patterns from functional MRI (fMRI) data. Previous studies have not found a clear benefit for non-linear (polynomial kernel) SVM versus linear one. Here, a more effective non-linear SVM using radial basis function (RBF) kernel is compared with linear SVM. Different from traditional studies which focused either merely on the evaluation of different types of SVM or the voxel selection methods, we aimed to investigate the overall performance of linear and RBF SVM for fMRI classification together with voxel selection schemes on classification accuracy and time-consuming. Methodology/Principal Findings Six different voxel selection methods were employed to decide which voxels of fMRI data would be included in SVM classifiers with linear and RBF kernels in classifying 4-category objects. Then the overall performances of voxel selection and classification methods were compared. Results showed that: (1) Voxel selection had an important impact on the classification accuracy of the classifiers: in a relative low dimensional feature space, RBF SVM outperformed linear SVM significantly; in a relative high dimensional space, linear SVM performed better than its counterpart; (2) Considering the classification accuracy and time-consuming holistically, linear SVM with relative more voxels as features and RBF SVM with small set of voxels (after PCA) could achieve the better accuracy and cost shorter time. Conclusions/Significance The present work provides the first empirical result of linear and RBF SVM in classification of fMRI data, combined with voxel selection methods. Based on the findings, if only classification accuracy was concerned, RBF SVM with appropriate small voxels and linear SVM with relative more voxels were two suggested solutions; if users concerned more about the computational time, RBF SVM with relative small set of voxels when part of the principal components were kept as features was a better choice.

Prediction of Progressive Mild Cognitive Impairment by Multi-Modal Neuroimaging Biomarkers

For patients with mild cognitive impairment (MCI), the likelihood of progression to probable Alzheimers disease (AD) is important not only for individual patient care, but also for the identification of participants in clinical trial, so as to provide early interventions. Biomarkers based on various neuroimaging modalities could offer complementary information regarding different aspects of disease progression. The current study adopted a weighted multi-modality sparse representation-based classification method to combine data from the Alzheimers Disease Neuroimaging Initiative (ADNI) database, from three imaging modalities: Volumetric magnetic resonance imaging (MRI), fluorodeoxyglucose (FDG) positron emission tomography (PET), and florbetapir PET. We included 117 normal controls (NC) and 110 MCI patients, 27 of whom progressed to AD within 36 months (pMCI), while the remaining 83 remained stable (sMCI) over the same time period. Modality-specific biomarkers were identified to distinguish MCI from NC and to predict pMCI among MCI. These included the hippocampus, amygdala, middle temporal and inferior temporal regions for MRI, the posterior cingulum, precentral, and postcentral regions for FDG-PET, and the hippocampus, amygdala, and putamen for florbetapir PET. Results indicated that FDG-PET may be a more effective modality in discriminating MCI from NC and in predicting pMCI than florbetapir PET and MRI. Combining modality-specific sensitive biomarkers from the three modalities boosted the discrimination accuracy of MCI from NC (76.7%) and the prediction accuracy of pMCI (82.5%) when compared with the best single-modality results (73.6% for MCI and 75.6% for pMCI with FDG-PET).

The Altered Triple Networks Interaction in Depression under Resting State Based on Graph Theory

The triple network model (Menon, 2011) has been proposed, which helps with finding a common framework for understanding the dysfunction in core neurocognitive network across multiple disorders. The alteration of the triple networks in the major depression disorder (MDD) is not clear. In our study, the altered interaction of the triple networks, which include default model network (DMN), central executive network (CEN), and salience network (SN), was examined in the MDD by graph theory method. The results showed that the connectivity degree of right anterior insula (rAI) significantly increased in MDD compared with healthy control (HC), and the connectivity degree between DMN and CEN significantly decreased in MDD. These results not only supported the proposal of the triple network model, but also prompted us to understand the dysfunction of neural mechanism in MDD.

Altered effective connectivity model in the default mode network between bipolar and unipolar depression based on resting-state fMRI

BACKGROUND Bipolar depression (BD) is characterized by alternating episodes of depression and mania. Patients who spend the majority of their time in episodes of depression rather than mania are often misdiagnosed with unipolar depression (UD) that only exhibits depressive episodes. It would be important to explore the construction of more objective biomarkers which can be used to more accurately differentiate BD and UD. METHODS The effective connectivity model of BD and UD in the default mode network (DMN) was constructed based on resting-state fMRI data of 17 BD (32.12±8.57 years old) and 17 UD (32.59±9.77 years old) patients using a linear non-Gaussian acyclic model (LiNGAM). The effective connectivity differences were obtained by conducting a permutation test. RESULTS The following connections were stronger in the BD group than in the UD group: medial prefrontal cortex (MPFC) →posterior cingulate cortex (PCC), right inferior parietal cortex (rIPC)→left hippocampus (lHC) and rIPC→right insula (rInsula). In contrast, the following connections were weak or unapparent in the BD group: MPFC→lHC, rHC→MPFC, rHC→rInsula and rInsula→lHC. LIMITATIONS First, the medication effect is a confounding factor. Second, as with most fMRI studies, the subjects׳ thoughts during imaging are difficult to control. CONCLUSIONS The brain regions in these altered connections, such as the HC, insula, MPFC and IPC, all play important roles in emotional processing, suggesting that these altered connections may be conducive to better distinguish between BD and UD.

Combining features from ERP components in single-trial EEG for discriminating four-category visual objects

Categorization of images containing visual objects can be successfully recognized using single-trial electroencephalograph (EEG) measured when subjects view images. Previous studies have shown that task-related information contained in event-related potential (ERP) components could discriminate two or three categories of object images. In this study, we investigated whether four categories of objects (human faces, buildings, cats and cars) could be mutually discriminated using single-trial EEG data. Here, the EEG waveforms acquired while subjects were viewing four categories of object images were segmented into several ERP components (P1, N1, P2a and P2b), and then Fisher linear discriminant analysis (Fisher-LDA) was used to classify EEG features extracted from ERP components. Firstly, we compared the classification results using features from single ERP components, and identified that the N1 component achieved the highest classification accuracies. Secondly, we discriminated four categories of objects using combining features from multiple ERP components, and showed that combination of ERP components improved four-category classification accuracies by utilizing the complementarity of discriminative information in ERP components. These findings confirmed that four categories of object images could be discriminated with single-trial EEG and could direct us to select effective EEG features for classifying visual objects.

An automated detection and correction method of EOG artifacts in EEG-based BCI

In BCI (Brain-Computer Interface) research community, most BCI research is focused on bioelectrical brain signals recorded by EEG (electroencephalography) as its noninvasive and thus readily available. While the EEG signal processing methods in EEG based BCI are appealing, they face substantial practical problems. Due to the limitation of EEG signal recording technology, physiological artifacts, especially those generated by eye (EOG, electrooculography), interfere with EEG, may change the characteristics of the neurological phenomena in EEG, and make those signal processing performs incompetently. Linear combination and regression is the most common used technique for removing ocular artifacts from EEG signals where a fraction of the EOG signal is subtracted from the EEG. One problem is that subtracting the EOG signal may also remove part of the EEG signal, for the EOG signal to be subtracted is also contaminated with the EEG signal. In this paper, a new EOG correction model is introduced for EOG artifacts, where the EEG contained in the EOG is considered, and thus avoid removing part of the EEG signal by subtracting the EOG signal. In order to apply this new model in online BCI signal processing, this paper adopts the AR (autoregressive) filtering model of the EEG activity to detect the EOG artifacts, only if it exists, the EEG correction method are performed. We test our methods in the BCI competition 2008 dataset IIa, our informal results indicate that EOG artifacts are well detected, and EOG is well removed from motor imagination related EEG signals.

Multimodal Classification of Mild Cognitive Impairment Based on Partial Least Squares

In recent years, increasing attention has been given to the identification of the conversion of mild cognitive impairment (MCI) to Alzheimers disease (AD). Brain neuroimaging techniques have been widely used to support the classification or prediction of MCI. The present study combined magnetic resonance imaging (MRI), 18F-fluorodeoxyglucose PET (FDG-PET), and 18F-florbetapir PET (florbetapir-PET) to discriminate MCI converters (MCI-c, individuals with MCI who convert to AD) from MCI non-converters (MCI-nc, individuals with MCI who have not converted to AD in the follow-up period) based on the partial least squares (PLS) method. Two types of PLS models (informed PLS and agnostic PLS) were built based on 64 MCI-c and 65 MCI-nc from the Alzheimers Disease Neuroimaging Initiative (ADNI) database. The results showed that the three-modality informed PLS model achieved better classification accuracy of 81.40%, sensitivity of 79.69%, and specificity of 83.08% compared with the single-modality model, and the three-modality agnostic PLS model also achieved better classification compared with the two-modality model. Moreover, combining the three modalities with clinical test score (ADAS-cog), the agnostic PLS model (independent data: florbetapir-PET; dependent data: FDG-PET and MRI) achieved optimal accuracy of 86.05%, sensitivity of 81.25%, and specificity of 90.77%. In addition, the comparison of PLS, support vector machine (SVM), and random forest (RF) showed greater diagnostic power of PLS. These results suggested that our multimodal PLS model has the potential to discriminate MCI-c from the MCI-nc and may therefore be helpful in the early diagnosis of AD.

The contribution of different frequency bands of fMRI data to the correlation with EEG alpha rhythm.

Alpha rhythm is a prominent EEG rhythm observed during resting state and is thought to be related to multiple cognitive processes. Previous simultaneous electroencephalography (EEG)/functional magnetic resonance imaging (fMRI) studies have demonstrated that alpha rhythm is associated with blood oxygen level dependent (BOLD) signals in several different functional networks. How these networks influence alpha rhythm respectively is unclear. The low-frequency oscillations (LFO) in spontaneous BOLD activity are thought to contribute to the local correlations in resting state. Recent studies suggested that either LFO or other components of fMRI can be further divided into sub-components on different frequency bands. We hypothesized that those BOLD sub-components characterized the contributions of different brain networks to alpha rhythm. To test this hypothesis, EEG and fMRI data were simultaneously recorded from 17 human subjects performing an eyes-close resting state experiment. EEG alpha rhythm was correlated with the filtered fMRI time courses at different frequency bands (0.01-0.08 Hz, 0.08-0.25 Hz, 0.01-0.027 Hz, 0.027-0.073 Hz, 0.073-0.198 Hz, and 0.198-0.25 Hz). The results demonstrated significant relations between alpha rhythm and the BOLD signals in the visual network and in the attention network at LFO band, especially at the very low frequency band (0.01-0.027 Hz).

A Triple Network Connectivity Study of Large-Scale Brain Systems in Cognitively Normal APOE4 Carriers

The triple network model, consisting of the central executive network (CEN), salience network (SN) and default mode network (DMN), has been recently employed to understand dysfunction in core networks across various disorders. Here we used the triple network model to investigate the large-scale brain networks in cognitively normal apolipoprotein e4 (APOE4) carriers who are at risk of Alzheimer’s disease (AD). To explore the functional connectivity for each of the three networks and the effective connectivity among them, we evaluated 17 cognitively normal individuals with a family history of AD and at least one copy of the APOE4 allele and compared the findings to those of 12 individuals who did not carry the APOE4 gene or have a family history of AD, using independent component analysis (ICA) and Bayesian network (BN) approach. Our findings indicated altered within-network connectivity that suggests future cognitive decline risk, and preserved between-network connectivity that may support their current preserved cognition in the cognitively normal APOE4 allele carriers. The study provides novel sights into our understanding of the risk factors for AD and their influence on the triple network model of major psychopathology.

A pooling-LiNGAM algorithm for effective connectivity analysis of fMRI data.

The Independent Component Analysis (ICA)—linear non-Gaussian acyclic model (LiNGAM), an algorithm that can be used to estimate the causal relationship among non-Gaussian distributed data, has the potential value to detect the effective connectivity of human brain areas. Under the assumptions that (a): the data generating process is linear, (b) there are no unobserved confounders, and (c) data have non-Gaussian distributions, LiNGAM can be used to discover the complete causal structure of data. Previous studies reveal that the algorithm could perform well when the data points being analyzed is relatively long. However, there are too few data points in most neuroimaging recordings, especially functional magnetic resonance imaging (fMRI), to allow the algorithm to converge. Smiths study speculates a method by pooling data points across subjects may be useful to address this issue (Smith et al., 2011). Thus, this study focus on validating Smiths proposal of pooling data points across subjects for the use of LiNGAM, and this method is named as pooling-LiNGAM (pLiNGAM). Using both simulated and real fMRI data, our current study demonstrates the feasibility and efficiency of the pLiNGAM on the effective connectivity estimation.