Computational Intelligence and Neuroscience

Computational Intelligence and Neuroscience / 2020 / Article
Special Issue

Advanced Techniques in Machine Learning for Neural Data Processing

View this Special Issue

Research Article | Open Access

Volume 2020 |Article ID 6405930 | https://doi.org/10.1155/2020/6405930

ZhiHong Chen, Tao Yan, ErLei Wang, Hong Jiang, YiQian Tang, Xi Yu, Jian Zhang, Chang Liu, "Detecting Abnormal Brain Regions in Schizophrenia Using Structural MRI via Machine Learning", Computational Intelligence and Neuroscience, vol. 2020, Article ID 6405930, 13 pages, 2020. https://doi.org/10.1155/2020/6405930

Detecting Abnormal Brain Regions in Schizophrenia Using Structural MRI via Machine Learning

Academic Editor: Horacio Rostro-Gonzalez
Received13 Oct 2019
Revised07 Jan 2020
Accepted16 Jan 2020
Published18 Mar 2020

Abstract

Utilizing neuroimaging and machine learning (ML) to differentiate schizophrenia (SZ) patients from normal controls (NCs) and for detecting abnormal brain regions in schizophrenia has several benefits and can provide a reference for the clinical diagnosis of schizophrenia. In this study, structural magnetic resonance images (sMRIs) from SZ patients and NCs were used for discriminative analysis. This study proposed an ML framework based on coarse-to-fine feature selection. The proposed framework used two-sample t-tests to extract the differences between groups first, then further eliminated the nonrelevant and redundant features with recursive feature elimination (RFE), and finally utilized the support vector machine (SVM) to learn the decision models with selected gray matter (GM) and white matter (WM) features. Previous studies have tended to report differences at the group level instead of at the individual level and cannot be widely applied. The method proposed in this study extends the diagnosis to the individual level and has a higher recognition rate than previous methods. The experimental results of this study demonstrate that the proposed framework distinguishes SZ patients from NCs, with the highest classification accuracy reaching over 85%. The identified biomarkers are also consistent with previous literature findings. As a universal method, the proposed framework can be extended to diagnose other diseases.

1. Introduction

Schizophrenia (SZ) is a group of major psychiatric diseases with unknown etiology. SZ has the highest prevalence of all mental illnesses and is very difficult to treat. Over the last few decades, many neuroimaging studies have demonstrated that schizophrenia is a disorder involving widespread abnormalities in the brain structure [15]. However, the specific mechanisms involved in producing these structural deficits remain incompletely understood. In recent years, it has been consistently reported that SZ patients have structural abnormalities in the brain, including the middle temporal gyrus, middle frontal gyrus, thalamus, and corpus callosum (CcSum) [68]. The brain structure location and neurobiological processes underlying these structural abnormalities are central to the pathophysiology of schizophrenia. Furthermore, alterations to the brain structure are linked to key psychotic symptoms (such as auditory hallucinations [9, 10], neurosensory deficits [11, 12], and social dysfunction [13, 14] in SZ).

At present, the diagnosis and monitoring of SZ mainly hinge on doctors’ judgment through patients’ clinical response, history, and neurological examination. The diagnosis and monitoring of SZ are heavily dependent on doctors’ clinical experience and related knowledge. In other words, this subjective judgment may add risk to the diagnosis and treatment of SZ. For a more accurate diagnosis, neuroimaging methods have been widely used to study brain morphology, which provides important information about possible pathophysiologic mechanisms [1518].

Due to its good contrast and high spatial resolution, structural magnetic resonance imaging (sMRI) has become one of the most popular neuroimaging modalities [1921]. Most existing research has investigated conventional statistical analysis methods to explore the differences between SZ patients and normal controls (NCs) based on group studies [15, 22, 23]. Despite the ability of conventional statistical analysis methods to detect some abnormal brain regions in SZ, they are univariate methods and often overlook the correlations among voxels, which often contain important characteristic information. Furthermore, conventional statistical analysis only considers differences among groups, and it is difficult to generalize the diagnosis to individual patients.

To overcome the drawbacks of conventional statistical analysis, machine learning (ML) techniques have been applied to analyze neuroimaging data. These techniques can extract stable structural or functional patterns from neuroimaging data and may potentially be useful for finding significant neuroimaging-based biomarkers. Currently, promising results have been reported for the classification of SZ patients and NCs [2426].

The most common feature of sMRI is the so-called brain tissue volume (obtained from voxel-based morphometry). However, the existence of too many irrelevant features can greatly degrade the classification accuracy, especially in neuroimaging studies. The preprocessed brain MRI may contain >100,000 nonzero voxels. In comparison, the sample size (number of subjects or observations) is often less than 1000 [27]. Thus, the number of features (voxels) greatly exceeds the number of observations (sample size). This issue is a common problem in machine learning studies and is known as the “curse of dimensionality” [2830]. The curse of dimensionality can lead to overfitting of the learned model. Therefore, choosing and utilizing appropriate feature selection methods can effectively improve the performance of the model.

For most supervised ML studies, the corresponding supervised feature selection method uses high-dimensional neuroimaging data and the required outcome labels (e.g., +1 treatment responders and −1 treatment nonresponders) to select relevant features and discard redundant features and noise [27]. More specifically, these techniques are subdivided into three categories [3032]: (1) “filter methods,” which use simple statistical measures (e.g., mean, variance, and correlation coefficients) to rank features according to their relevance in detecting group-level differences, such as s t-tests, analysis of variance (ANOVA), and Pearson correlation coefficients; (2) “wrapper methods,” which use a cost function to optimize the machine learning model and rank features in terms of their relevance; and (3) “embedded methods,” which select relevant features as “part” of the machine learning process by enforcing certain “penalties” on the machine learning model to yield a small subset of relevant features.

A recent study [33] utilized a support vector machine (SVM) to learn the decision model to classify Alzheimer’s disease patients and normal controls, which achieved an area under the curve (AUC) of over 88.82%. Another study [8] used cortical thickness in conjunction with surface area in schizophrenia patients to perform discriminative analysis and obtained an accuracy of 85.0%. Though the good performance with machine learning produced an excessive number of features (voxels), there is a risk of overfitting [34, 35]. To overcome this problem, a large number of feature selection methods have been proposed. One study [36] selected discriminative features using Fisher’s criterion to train the SVM model. As a result, the classification accuracy reached 76.25% for identifying bipolar disorder patients from normal controls. Another study [37] detected first episode psychosis, which utilized principal component analysis (PCA) to reduce the number of nonrelevant features in cortical thickness and gray matter volume and then applied deep neural networks (DNNs) to construct the classification model. The authors achieved a classification accuracy of over 70.5%. However, most of these studies have analyzed only gray matter (GM). In fact, several studies [17, 38, 39] have demonstrated that there is a nonnegligible change in white matter (WM) in SZ and that it is also necessary to analyze WM.

As an effective feature selection algorithm, recursive feature elimination (RFE) evaluates the contribution of each feature and then eliminates the smallest contribution features iteratively [4042]. In this study, a machine learning framework based on coarse-to-fine feature selection is proposed. The framework first uses two-sample t-tests to roughly select features and then eliminates the nonrelevant and redundant features via RFE. Finally, the SVM is utilized to learn the decision models for WM and GM separately. The experimental results demonstrate that the proposed method is able to differentiate SZ patients from NCs with a maximum accuracy of approximately 85% and can find biomarkers of SZ that are consistent with those found in previous studies, including the left and right middle temporal gyrus, right middle frontal gyrus, thalamus, corpus callosum, fusiform gyrus, occipital lobe, cuneus, postcentral gyrus, and cerebellum.

The contributions of our work include the following:(1)We developed a machine learning framework to differentiate SZ patients from NCs. The proposed machine learning framework adopts a coarse-to-fine approach to roughly reduce the dimensionality of features with two-sample t-tests and then further with RFE. Hierarchical feature selection is helpful for preserving informative features and eliminating redundant ones. Furthermore, coarse-to-fine feature selection is easy to use to identify biomarkers of schizophrenia (abnormal brain regions). The proposed machine learning framework does not apply to schizophrenia only. It can also be generalized to other diseases to classify patients and NCs based on sMRI.(2)The experimental results demonstrate that the proposed method achieves a better classification performance than other methods. Furthermore, the identified biomarkers are consistent with the findings of previous related research works.(3)Previous research works have mainly focused on gray matter and have seldom investigated white matter in schizophrenia patients. This study analyzes gray matter and white matter separately and finds that white matter has a better discriminative ability than gray matter, which provides a reference for clinical diagnosis.

2. Materials and Methods

2.1. Subjects and MRI Data Acquisition

The imaging data and phenotypic information used in this study were obtained from the Centers for Biomedical Research Excellence (COBRE) dataset, which was collected and shared by the Mind Research Network and The University of New Mexico (http://fcon_1000.projects.nitrc.org/indi/retro/cobre.html). To reduce the impact of different subtypes of SZ, we chose only paranoid schizophrenia from the dataset. Paranoid schizophrenia is the most common type of SZ and has a slower course of disease development, a later neurodegenerative onset time, and a better curative effect [43, 44].

In this study, we selected 34 paranoid schizophrenia patients and 34 normal controls from the dataset. The selected subjects were right-handed and were aged between 20 and 60. All subjects were examined and excluded if they had a history of a neurological disorder, a history of mental retardation, a history of severe head trauma with more than 5 minutes of loss of consciousness, or a history of substance abuse or dependence within the last 12 months. Diagnostic information was gathered using the Structured Clinical Interview for DSM Disorders (SCID). The demographics are reported in Table 1.


SZNC

Gender (M/F)27/723/11
Age36.85 ± 10.9139.53 ± 10.59

Note: SZ, schizophrenic; NC, normal control; M, male; F, female.

All sMRI data were acquired with a multiecho MPRAGE (MEMPR) sequence. The parameters used were a repetition time (TR) of 2530 ms; echo times (TEs) of 1.64, 3.5, 5.36, 7.22, and 9.08 ms; an inversion time (TI) of 900 ms; an FOV (field of view) of 256 × 256 mm; a matrix of 256 × 256 × 176; a flip angle of 7°; a voxel size of 1 × 1 × 1 mm; a slab thickness of 176 mm; a number of echoes of 5; and a total scan time of 6 min.

2.2. Preprocessing

sMRI data were analyzed with the Statistical Parametric Mapping (SPM) software package SPM8 (Wellcome Department of Imaging Neuroscience, London, UK; http://www.fil.ion.ucl.ac.uk/spm) using the voxel-based morphometry (VBM) [45, 46] protocol. First, spatial normalization of all 3D volumes with the T1 template was provided by SPM8 for bias correction (removal of positional and volume differences). Second, each T1-weighted MRI was segmented into three tissue probability maps (TPMs), including GM, WM, and cerebrospinal fluid (CSF). Third, the tissue volume was obtained by modulating the segmented tissue maps. Finally, a Gaussian kernel with a 6 mm isotropic full width at half maximum was employed for spatial smoothing.

2.3. Machine Learning Framework

After the preprocessing step, feature selection based on the coarse-to-fine approach was conducted to reduce the dimensionality of the features. First, two-sample t-tests were conducted to roughly select features, and then, RFE was used to further eliminate nonrelevant and redundant features. Lastly, a linear SVM classifier was trained to classify SZ patients and NCs. The workflow of the proposed machine learning framework is shown in Figure 1.

2.4. Feature Selection

To obtain a good classification performance, two-sample t-tests were used to perform a rough preliminary selection in this paper. Then, RFE was used to further select discriminative features.

2.4.1. Two-Sample t-Tests

Due to the large amount of redundant information in sMRI, two-sample t-tests were used to initially screen the voxels. As a classical statistical analysis method, two-sample t-tests can extract significant differences between groups by computing the statistical significance value. Suppose and represent the means of a feature of the two groups. and denote the corresponding variances, and the significant differences between groups on this feature can be calculated as follows:where and denote the sample sizes. The ability of a feature to distinguish between two groups is evaluated by the absolute value of . The greater the absolute value of , the more discriminative the feature.

2.4.2. Recursive Feature Elimination

RFE [47] is a greedy method for ranking all features to obtain an optimal feature subset for classification. To perform this ranking, RFE trains a machine learning model (e.g., linear support vector machine or relevance vector machine), then ranks all features in terms of some specific ranking criteria, and finally removes the features with the smallest rankings. The procedure is repeated until all features are removed. Since RFE can eliminate a fixed quantity or percentage of features depending on the user’s requirements and has a strong ability to explain differences, it has been popular in neuroimaging studies [4042].

Currently, most studies [4851] have combined RFE with the SVM to perform feature selection. The SVM is presently one of the best-known classification techniques and has computational advantages over other classification methods, and many previous studies [5256] have proven that the linear SVM performs well in small sample datasets. To allow the classifier to generalize unseen data well and to avoid overfitting problems, we introduced the SVM soft margin classifier.

Taking the soft margin SVM as an example, assuming input training samples , each sample has features (i.e., , ) and corresponding labels ; the decision function is formulated aswhere is the weights of features, is a nonzero penalty coefficient that controls the trade-off between the training error and the margin, and is called slack variables that are associated with the misclassified samples.

Since the above optimization problem is difficult to solve, it can be rewritten as a dual problem using a Lagrangian multiplier method as follows:where corresponds to the weights of observation samples. The observation samples with nonzero weights represent support vectors. Consequently, the weights of features or voxels are calculated as

To evaluate the contribution of each feature, the weight is ranked based on its squared value . Finally, the lowest ranking feature is removed from feature sets :

Subsequently, the above process is iterated until a termination criterion is reached or until the feature set is empty. Then, each feature corresponds to a weight, which expresses the importance of the feature. Finally, a user-defined ratio (e.g., 2%) is applied to remove the lower ranking features. The RFE process is shown in Figure 2.

2.5. Performance Evaluation

To assess the performance of the proposed method robustly, leave-one-out cross validation (LOO) was applied in this study. Suppose there are samples of SZ patients and samples of NCs; of them, samples of SZ patients and samples of NCs are correctly classified. Five performance measures (accuracy, sensitivity, specificity, F2 measure, and G mean) were used to evaluate the performance:where , , , and . represents the proportion of SZ patients predicted correctly, and represents the proportion of normal controls predicted correctly. , , and are defined as the harmonic mean of and precision , of which and are computed with the same weights of and precision and is computed with higher weights of .

3. Results

To verify the performance of the coarse-to-fine feature selection proposed in this study, we compared it with six other machine learning methods: (1) directly using SVM to classify (SVM), (2) using two-sample t-tests to select features and SVM to classify (2T + SVM), (3) using RFE to select features and SVM to classify (RFE + SVM), (4) using principal component analysis (PCA) and SVM to classify (PCA + SVM), (5) using independent component analysis (ICA) and SVM to classify (ICA + SVM), and (6) using tree-based feature selection and SVM to classify (TBFS + SVM). Furthermore, to test the performance of the proposed machine learning framework, we compared it with other frameworks that select features roughly based on 2T, perform PCA (ICA, TBFS), and then apply SVM for classification (2T + PCA + SVM, 2T + ICA + SVM, 2T + TBFS + SVM). Finally, we analyzed the biomarkers of SZ using coarse-to-fine feature selection.

3.1. Parameter Setting

A linear SVM was applied in this study, and previous studies [52, 53] proved that the linear SVM works better on small sample datasets. The value of the penalty coefficient “C” was set to 1.0 because many experimental tests have shown that a value of 1.0 can obtain a satisfactory discrimination performance.

The number of retained features has a significant impact on the results when using RFE. We tested the effect of different numbers of retained features on the results. The experimental results showed that retaining 40% and 14% of the voxels for GM and WM, respectively, and using an elimination ratio of voxels each round of 5% achieved the best performance.

3.2. Two-Sample t-Tests

To test the performance of different values, two-sample t-tests were performed with three different values (<0.05 [5759], <0.01 [60], and <0.001 [61]). The cluster-size value was set to 50 [6265], and three differentiated tissue maps of GM and WM were obtained and are shown in Figures 3 and 4. From the figures, we can see that as the values decrease, the region selected (red parts) becomes smaller. Consequently, the smaller the value, the more the information being filtered, which may result in the removal of useful features. Therefore, the criterion was adopted in our coarse-to-fine feature selection algorithm.

3.3. Classification Performance

The results of different methods based on GM and WM are shown in Tables 2 and 3, respectively.


MethodACCSNSPGMDMF2M

SVM0.62770.69230.57110.62880.61670.6339
2T + SVM0.68290.72260.64310.68170.67410.6812
RFE + SVM0.73260.84270.65330.74200.73210.7699
PCA + SVM0.70570.77120.65210.70920.70290.7103
ICA + SVM0.68060.71140.68620.69870.65710.6968
TBFS + SVM0.71180.82000.62670.71690.71550.7327
2T + PCA + SVM0.79810.81930.77040.79450.78960.7922
2T + ICA + SVM0.75040.77270.71110.74130.73590.7394
2T + TBFS + SVM0.77940.80090.74840.77420.76580.7701
Ours0.79620.81420.77770.79570.78510.7896


MethodACCSNSPGMDMF2M

SVM0.56890.56740.57160.56950.56210.5665
2T + SVM0.78320.78180.78710.78440.77980.7788
RFE + SVM0.64270.67000.63000.64970.64180.6469
PCA + SVM0.61720.59680.63700.61660.61010.5996
ICA + SVM0.57740.57960.58300.58130.57210.5740
TBFS + SVM0.58790.55470.62110.58670.56620.5593
2T + PCA + SVM0.83720.84740.83230.83980.83380.8375
2T + ICA + SVM0.79920.79920.80370.80140.79520.7978
2T + TBFS + SVM0.81970.79260.83510.81360.80800.7965
Ours0.85270.85870.85080.85470.84970.8532

From Table 2, we can see that feature selection can achieve a better classification performance than using SVM directly for GM. When using only one feature selection algorithm, such as 2T + SVM, RFE + SVM, or PCA + SVM, RFE + SVM can achieve the best performance, which shows that RFE has a better feature selection ability than other methods. When the coarse-to-fine approach is used, such as 2T + RFE + SVM, 2T + PCA + SVM achieves the best performance and reaches an accuracy of 79.81%. The proposed 2T + RFE + SVM method achieves a similar performance (accuracy of 79.62%).

From Table 3, when only one feature selection algorithm is used for WM, all five feature selection methods (2T + SVM, RFE + SVM, PCA + SVM, ICA + SVM, and TBFS-SVM) have a better classification performance than using SVM directly. However, the performances of RFE + SVM, PCA + SVM, ICA + SVM, and TBFS + SVM are worse than that of 2T + SVM for WM. The reason for this result may be that WM has more redundant features, and RFE, PCA, and other methods did not identify useful features from many irrelevant voxels. In contrast, as a traditional statistical analysis, it is easy to explore significant differences based on prior knowledge about different groups. Four coarse-to-fine frameworks perform better than using single feature selection, and the proposed 2T + RFE + SVM method achieves the best performance and reaches an accuracy of over 85% for WM.

According to the above experiments, we can see that coarse-to-fine feature selection selects more discriminative features. Using the same coarse-to-fine machine learning framework, RFE can achieve a better performance than the others, and the receiver-operating characteristic (ROC) curves of WM and GM with the proposed ML framework are shown in Figure 5. The AUC represents the performance of the classification experiment. The larger the AUC, the better the performance. It can be seen that the AUC for WM is better than that for GM.

3.4. Identification of Abnormal Brain Regions

The discriminative brain regions (biomarkers) of GM and WM selected by the proposed method are illustrated in Figure 6. We selected clusters as biomarkers when the cluster size was ≥50. For GM, 14 brain regions were detected. For WM, 24 brain regions were detected. Detailed brain biomarkers are shown in Tables 4 and 5. From Table 4, we can see that the GM brain regions that were detected included the cerebellum, fusiform gyrus, temporal lobe, occipital lobe, frontal lobe, right supramarginal gyrus, angular gyrus, and postcentral gyrus. From Table 5, we found that the WM brain regions that were detected included the cerebellum, fusiform gyrus, temporal lobe, occipital lobe, frontal lobe, lentiform nucleus, thalamus, corpus callosum, cuneus, subgyral, and postcentral gyrus. The selected abnormal brain regions were similar in GM and WM. This finding shows that SZ can cause changes in specific brain regions, and these regions are also considered SZ biomarkers.


LocationRegion of interest (ROI)Size of clusterMNI coordinates value
AALBAXYZ

lCbe91055167−6−55.5−450.05
rCbe9106889−54−450.05
FFG.R562025131.5−8−41.50.05
rCbeCru1929634.5−63−390.05
lCbeCrul91168−36−53.5−370.05
lCbe69987−36−52−41.50.05
lCbe4-59761−18−42−28.50.05
ITG.R90206355.5−30−300.05
MOG.L5117127−26.5−96.500.05
SOG.L491762−22−96.515.50.05
MTG.R863731145.5−6614.50.05
MTG.L853769−48−61.54.50.05
IFGtriang.R14486040.519.5210.05
SMG.R64487751−48270.05
ANG.R66406936−56370.05
PoCG.L57387−21−34.558.50.05


LocationRegion of interest (ROI)Size of clusterMNI coordinates value
AALBAJHUXYZ

lCbe810376−30−58.5−52.50.05
rCbe81047937.5−50−40.50.05
rCbe9106537.5−50−450.05
FFG.R562025636−12.5−34.50.05
ITG.R90367832.5−1−41.50.05
lCbe910567−9−57−43.50.05
FFG.L552094−32.5−10.5−300.05
FFG.R56186630−81.5−100.05
ORBinf.L154761−37.539−150.05
PUT.R74483313631.59−60.05
SOG.R50177218−102−4.50.05
THA.R7823916.5−30−1.50.05
PUT.L734834127−28.5040.05
SOG.L491792−12−10090.05
IFGoperc.L114880−4813.5150.05
CcSum4731.51516.50.05
MFG.R8487537.53421.50.05
CUN.L451870−7.5−79.522.50.05
PoCG.R58438663−13.5240.05
IPL.L614070−46.5−3934.50.05
Subgyral26160−21637.50.05
PoCG.L5753−19.5−43.543.50.05
PCUN.R685413.5−49.5480.05

4. Discussion

Previous group-level statistical analysis of neuroimaging data uncovered some neuroanatomical and functional differences between SZ patients and NCs [6668]. Nevertheless, those findings had limited clinical applications. Machine learning can be adopted for single subject prediction and has shown significant potential for disease diagnosis [6972] at the individual level.

To verify the performance of the proposed method, we performed the following experiments: (1) We used the SVM directly on smoothed voxels. (2) We used single feature selection and SVM, such as 2T + SVM, RFE + SVM, PCA + SVM, ICA + SVM, and TBFS + SVM. (3) We used coarse-to-fine feature selection methods for classification, such as 2T + PCA + SVM, 2T + ICA + SVM, 2T + TBFS + SVM, and 2T + RFE + SVM. The experiments illustrated the following: (1) SVM classification with feature selection is better than direct SVM classification, which means that the feature selection method can discard redundant features and extract useful information. (2) Among the single feature selection frameworks, 2T + SVM achieved good results for WM, but 2T + SVM required prior knowledge about the groups, which may lead to the poor generalizability of the learning model. Furthermore, using the same framework (RFE + SVM, PCA + SVM, etc. vs. 2T + RFE + SVM, 2T + PCA + SVM, etc.), the frameworks for RFE were synthetically better, which indicated that the features extracted by RFE were more efficient. (3) The performance of the coarse-to-fine framework was better than that of single feature selection methods, which indicates that selecting features hierarchically can filter features more effectively.

Structural abnormalities have been repeatedly demonstrated in SZ patients compared to NCs in previous MRI studies [7375]. However, most research has focused on DTI to analyze structural changes in SZ and has rarely found structural differences on sMRI. In the current study, the GM and WM features selected by the proposed method provided discriminatory information about anatomically abnormal patterns in schizophrenia. The proposed method in the current study revealed sensitive and accurate information about anatomically abnormal patterns in the frontal lobe, postcentral gyrus, corpus callosum, and cuneus, especially in the thalamus, fusiform gyrus, temporal lobe, and cerebellum. These identified abnormal biomarkers were consistent with those found in the literature [8, 7679]. Some brain regions were found in both GM and WM, including the cerebellum, fusiform gyrus, temporal lobe, occipital lobe, and frontal lobe, which showed that SZ could indeed cause structural changes in these brain regions. Many previous studies have reported that the concentrations of some substances (such as dopamine) change in the cingulate cortex and amygdale in SZ, but we did not detect structural changes in these areas. The possible reason is that changes in the dopamine concentration do not cause structural abnormalities.

This study has several limitations. First, a small sample size is a common pitfall in most similar studies. To improve the generalizability and clinical applicability of machine learning methods, more samples need to be collected in future studies. Second, only sMRI data were investigated. Multimodal data, such as fMRI, DTI, and brain connectivity data, remain to be explored to provide either complementary or additional information for accurate recognition and lateralization in SZ. Finally, we identified the discriminative regions based on the AAL, BA, and JHU atlas, with the potential drawback that some atlas areas might be too large or unspecific to detect group differences.

5. Conclusion

In this study, a coarse-to-fine ML framework was investigated to differentiate SZ patients from NCs and to detect biomarkers of SZ using sMRI. The experiments demonstrated that feature selection algorithms can effectively improve the classification performance. The use of coarse-to-fine feature selection extracted more effective information and significantly improved the classification accuracy. The experiments also indicated that the classification performance of WM was significantly better than that of GM. Therefore, it can be concluded that SZ has a greater impact on WM. This conclusion is consistent with previous findings. Furthermore, the proposed coarse-to-fine feature selection effectively located abnormal brain regions, which provides a helpful aid for the clinical diagnosis of SZ. As a universal method, the proposed framework can be extended to diagnose other diseases.

Data Availability

The datasets used in the current study can be obtained from the Centers for Biomedical Research Excellence (COBRE) at http://fcon_1000.projects.nitrc.org/indi/retro/cobre.html.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

Authors’ Contributions

ZhiHong Chen, Tao Yan, and ErLei Wang contributed equally to this study.

Acknowledgments

This work was supported by the National Natural Science Funds of China (Grant no. 61502059), the China Postdoctoral Science Foundation (Grant no. 2016M592656), and Sichuan Science and Technology Program (Grant no. 2018JY0272). The authors are grateful to other members of the Key Laboratory of Pattern Recognition and Intelligent Information Processing in Sichuan who offered assistance during this study.

References

  1. E. Schwarz, N. T. Doan, G. Pergola et al., “Reproducible grey matter patterns index a multivariate, global alteration of brain structure in schizophrenia and bipolar disorder,” Translational Psychiatry, vol. 9, no. 12, pp. 1–13, 2019. View at: Publisher Site | Google Scholar
  2. S. Shahab, B. H. Mulsant, M. L. Levesque et al., “Brain structure, cognition, and brain age in schizophrenia, bipolar disorder, and healthy controls,” Neuropsychopharmacology, vol. 44, no. 5, pp. 898–906, 2019. View at: Publisher Site | Google Scholar
  3. F. Fan, H. Xiang, S. Tan et al., “Subcortical structures and cognitive dysfunction in first episode schizophrenia,” Psychiatry Research: Neuroimaging, vol. 286, pp. 69–75, 2019. View at: Publisher Site | Google Scholar
  4. C. Chiapponi, P. De Rossi, F. Piras, T. Gili, and G. Spalletta, “Brain morphometry: schizophrenia,” Neuromethods: Brain Morphometry, Springer, Berlin, Germany, 2018. View at: Publisher Site | Google Scholar
  5. S. Fritze, A. L. Bertolino, K. M. Kubera et al., “Differential contributions of brainstem structures to neurological soft signs in first- and multiple-episode schizophrenia spectrum disorders,” Schizophrenia Research, vol. 210, pp. 101–106, 2019. View at: Publisher Site | Google Scholar
  6. G. Tarcijonas and D. K. Sarpal, “Neuroimaging markers of antipsychotic treatment response in schizophrenia: an overview of magnetic resonance imaging studies,” Neurobiology of Disease, vol. 131, Article ID 104209, 2019. View at: Publisher Site | Google Scholar
  7. M. Latha and G. Kavitha, “Segmentation and texture analysis of structural biomarkers using neighborhood-clustering-based level set in MRI of the schizophrenic brain,” Magnetic Resonance Materials in Physics, Biology and Medicine, vol. 31, no. 4, pp. 483–499, 2018. View at: Publisher Site | Google Scholar
  8. Y. Xiao, Z. Yan, Y. Zhao et al., “Support vector machine-based classification of first episode drug-naïve schizophrenia patients and healthy controls using structural MRI,” Schizophrenia Research, vol. 214, pp. 11–17, 2019. View at: Publisher Site | Google Scholar
  9. O. P. du Sert, S. Potvin, O. Lipp et al., “Virtual reality therapy for refractory auditory verbal hallucinations in schizophrenia: a pilot clinical trial,” Schizophrenia Research, vol. 197, pp. 176–181, 2018. View at: Publisher Site | Google Scholar
  10. D. Kimhy, M. M. Wall, M. C. Hansen et al., “Autonomic regulation and auditory hallucinations in individuals with schizophrenia: an experience sampling study,” Schizophrenia Bulletin, vol. 43, no. 4, pp. 754–763, 2017. View at: Publisher Site | Google Scholar
  11. A. F. T. Arnsten, R. R. Girgis, D. L. Gray, and R. B. Mailman, “Novel dopamine therapeutics for cognitive deficits in schizophrenia,” Biological Psychiatry, vol. 81, no. 1, pp. 67–77, 2017. View at: Publisher Site | Google Scholar
  12. Z.-Y. Yang, D.-J. Xie, Y.-M. Zou et al., “Prospection deficits in schizophrenia: evidence from clinical and subclinical samples,” Journal of Abnormal Psychology, vol. 127, no. 7, pp. 710–721, 2018. View at: Publisher Site | Google Scholar
  13. D. Mucheru, M.-C. Hanlon, L. Campbell et al., “Social dysfunction and diet outcomes in people with psychosis,” Nutrients, vol. 9, no. 80, pp. 1–14, 2017. View at: Publisher Site | Google Scholar
  14. R. Langdon, M. Connors, and E. Connaughton, “Indirect task instructions better reveal theory-of-mind impairment, independent of executive dysfunction, in schizophrenia,” Psychiatry Research, vol. 256, pp. 342–344, 2017. View at: Publisher Site | Google Scholar
  15. N. Yao, A. M. Winkler, J. Barrett et al., “Inferring pathobiology from structural MRI in schizophrenia and bipolar disorder: modeling head motion and neuroanatomical specificity,” Human Brain Mapping, vol. 38, no. 8, pp. 3757–3770, 2017. View at: Publisher Site | Google Scholar
  16. N. D. Orlov, O. O’Daly, D. K. Tracy et al., “Stimulating thought: a functional MRI study of transcranial direct current stimulation in schizophrenia,” Brain, vol. 140, no. 9, pp. 2490–2497, 2017. View at: Publisher Site | Google Scholar
  17. S. O’Donoghue, L. Holleran, D. M. Cannon et al., “Anatomical dysconnectivity in bipolar disorder compared with schizophrenia: a selective review of structural network analyses using diffusion MRI,” Journal of Affective Disorders, vol. 209, pp. 217–228, 2017. View at: Publisher Site | Google Scholar
  18. L. Palaniyappan, O. Hodgson, V. Balain, S. Iwabuchi, P. Gowland, and P. Liddle, “Structural covariance and cortical reorganisation in schizophrenia: a MRI-based morphometric study,” Psychological Medicine, vol. 49, no. 3, pp. 412–420, 2019. View at: Publisher Site | Google Scholar
  19. S. Meyer, K. Mueller, K. Stuke et al., “Predicting behavioral variant frontotemporal dementia with pattern classification in multi-center structural MRI data,” NeuroImage: Clinical, vol. 14, pp. 656–662, 2017. View at: Publisher Site | Google Scholar
  20. P. M. Ferraro, F. Agosta, N. Riva et al., “Multimodal structural MRI in the diagnosis of motor neuron diseases,” NeuroImage: Clinical, vol. 16, pp. 240–247, 2017. View at: Publisher Site | Google Scholar
  21. L. Ferschmann, A. M. Fjell, M. E. Vollrath, H. Grydeland, K. B. Walhovd, and C. K. Tamnes, “Personality traits are associated with cortical development across adolescence: a longitudinal structural MRI study,” Child Development, vol. 89, no. 3, pp. 811–822, 2018. View at: Publisher Site | Google Scholar
  22. M. Chang, F. Y. Womer, E. K. Edmiston et al., “Neurobiological commonalities and distinctions among three major psychiatric diagnostic categories: a structural MRI study,” Schizophrenia Bulletin, vol. 44, no. 1, pp. 65–74, 2017. View at: Publisher Site | Google Scholar
  23. E. F. Barry, L. D. Vanes, D. S. Andrews et al., “Mapping cortical surface features in treatment resistant schizophrenia with in vivo structural MRI,” Psychiatry Research, vol. 274, pp. 335–344, 2019. View at: Publisher Site | Google Scholar
  24. M. L. Birnbaum, S. K. Ernala, A. F. Rizvi et al., “A collaborative approach to identifying social media markers of schizophrenia by employing machine learning and clinical appraisals,” Journal of Medical Internet Research, vol. 19, Article ID e289, pp. 1–11, 2017. View at: Publisher Site | Google Scholar
  25. N. Honnorat, A. Dong, E. Meisenzahl-Lechner et al., “Neuroanatomical heterogeneity of schizophrenia revealed by semi-supervised machine learning methods,” Schizophrenia Research, vol. 214, pp. 43–50, 2019. View at: Publisher Site | Google Scholar
  26. H. G. Schnack, M. Nieuwenhuis, N. E. M. van Haren et al., “Can structural MRI aid in clinical classification? A machine learning study in two independent samples of patients with schizophrenia, bipolar disorder and healthy subjects,” Neuroimage, vol. 84, pp. 299–306, 2014. View at: Publisher Site | Google Scholar
  27. B. Mwangi, T. S. Tian, and J. C. Soares, “A review of feature reduction techniques in neuroimaging,” Neuroinformatics, vol. 12, no. 2, pp. 229–244, 2014. View at: Publisher Site | Google Scholar
  28. R. Bellman and R. E. Bellman, Adaptive Control Processes: A Guided Tour, Princeton University Press, Princeton, NJ, USA, 1961.
  29. R. Kohavi and G. H. John, “Wrappers for feature subset selection,” Artificial Intelligence, vol. 97, no. 1-2, pp. 273–324, 1997. View at: Publisher Site | Google Scholar
  30. J. Hua, W. D. Tembe, and E. R. Dougherty, “Performance of feature-selection methods in the classification of high-dimension data,” Pattern Recognition, vol. 42, no. 3, pp. 409–424, 2009. View at: Publisher Site | Google Scholar
  31. I. Guyon and A. Elisseeff, “An introduction to variable and feature selection,” Journal of Machine Learning Research, vol. 3, pp. 1157–1182, 2003. View at: Google Scholar
  32. Y. Saeys, I. Inza, and P. Larranaga, “A review of feature selection techniques in bioinformatics,” Bioinformatics, vol. 23, no. 19, pp. 2507–2517, 2007. View at: Publisher Site | Google Scholar
  33. K. R. Kruthika, H. Rajeswari, A. Pai, and H. D. Maheshappa, “Classification of alzheimer and MCI phenotypes on MRI data using SVM,” in Proceedings of the International Symposium on Signal Processing and Intelligent Recognition Systems, pp. 263–275, Bangalore, India, September 2018. View at: Google Scholar
  34. S. Yeom, I. Giacomelli, M. Fredrikson et al., “Privacy risk in machine learning: analyzing the connection to overfitting,” in Proceedings of the 2018 IEEE 31st Computer Security Foundations Symposium (CSF), pp. 268–282, Oxford, UK, July 2018. View at: Publisher Site | Google Scholar
  35. M. Belkin, D. J. Hsu, and P. Mitra, “Overfitting or perfect fitting? risk bounds for classification and regression rules that interpolate,” in Proceedings of the Advances in Neural Information Processing Systems, pp. 2300–2311, Montreal, Canada, December 2018. View at: Google Scholar
  36. O. Cigdem, R. Soyak, B. Aydeniz et al., “Classification of healthy siblings of bipolar disorder patients from healthy controls using MRI,” in Proceedings of the 2019 Medical Technologies Congress (TIPTEKNO), pp. 1–4, Izmir, Turkey, October 2019. View at: Publisher Site | Google Scholar
  37. S. Vieira, Q.-Y. Gong, W. H. Pinaya et al., “Using machine learning and structural neuroimaging to detect first episode psychosis: reconsidering the evidence,” Schizophrenia Bulletin, vol. 46, no. 1, pp. 17–26, 2020. View at: Publisher Site | Google Scholar
  38. O. Pasternak, C.-F. Westin, B. Dahlben, S. Bouix, and M. Kubicki, “The extent of diffusion MRI markers of neuroinflammation and white matter deterioration in chronic schizophrenia,” Schizophrenia Research, vol. 161, no. 1, pp. 113–118, 2015. View at: Publisher Site | Google Scholar
  39. Y. Jiang, C. Luo, X. Li et al., “White-matter functional networks changes in patients with schizophrenia,” NeuroImage, vol. 190, pp. 172–181, 2019. View at: Publisher Site | Google Scholar
  40. C. Lai, S. Guo, L. Cheng et al., “A comparative study of feature selection methods for the discriminative analysis of temporal lobe epilepsy,” Frontiers in Neurology, vol. 8, pp. 1–13, 2017. View at: Publisher Site | Google Scholar
  41. V. Wottschel, D. T. Chard, C. Enzinger et al., “SVM recursive feature elimination analyses of structural brain MRI predicts near-term relapses in patients with clinically isolated syndromes suggestive of multiple sclerosis,” NeuroImage: Clinical, vol. 24, Article ID 102011, 2019. View at: Publisher Site | Google Scholar
  42. D. Domashenko, M. Manko, A. Popov et al., “Feature ranking for mild cognitive impairment and Alzheimer’s disease diagnosis,” in Proceedings of the 2017 Signal Processing Symposium (SPSympo), pp. 1–4, Jachranka, Poland, September 2017. View at: Publisher Site | Google Scholar
  43. H. Huang, C. Shu, J. Chen et al., “Altered corticostriatal pathway in first-episode paranoid schizophrenia: resting-state functional and causal connectivity analyses,” Psychiatry Research: Neuroimaging, vol. 272, pp. 38–45, 2018. View at: Publisher Site | Google Scholar
  44. P. C. Jariwala, N. D. Shah, K. R. Dave et al., “Delusional misidentification syndromes in patients of paranoid schizophrenia: case series and review,” Dementia, vol. 3, no. 2, pp. 125–129, 2017. View at: Google Scholar
  45. J. Ashburner and K. J. Friston, “Voxel-based morphometry-the methods,” Neuroimage, vol. 11, no. 6, pp. 805–821, 2000. View at: Publisher Site | Google Scholar
  46. A. Mechelli, C. Price, K. Friston, and J. Ashburner, “Voxel-based morphometry of the human brain: methods and applications,” Current Medical Imaging Reviews, vol. 1, no. 2, pp. 105–113, 2005. View at: Publisher Site | Google Scholar
  47. I. Guyon, J. Weston, S. Barnhill, and V. Vapnik, “Gene selection for cancer classification using support vector machines,” Machine Learning, vol. 46, no. 1–3, pp. 389–422, 2002. View at: Publisher Site | Google Scholar
  48. C. Zhang, Y. Zhou, J. Guo, G. Wang, and X. Wang, “Research on classification method of high-dimensional class-imbalanced datasets based on SVM,” International Journal of Machine Learning and Cybernetics, vol. 10, no. 7, pp. 1765–1778, 2019. View at: Publisher Site | Google Scholar
  49. S. Huang, N. Cai, P. P. Pacheco et al., “Applications of support vector machine (SVM) learning in cancer genomics,” Cancer Genomics-Proteomics, vol. 15, pp. 41–51, 2018. View at: Publisher Site | Google Scholar
  50. Z. Shao, S. Yang, F. Gao, K. Zhou, and P. Lin, “A new electricity price prediction strategy using mutual information-based SVM-RFE classification,” Renewable and Sustainable Energy Reviews, vol. 70, pp. 330–341, 2017. View at: Publisher Site | Google Scholar
  51. S. Sahran, D. Albashish, A. Abdullah, N. A. Shukor, and S. Hayati Md Pauzi, “Absolute cosine-based SVM-RFE feature selection method for prostate histopathological grading,” Artificial Intelligence in Medicine, vol. 87, pp. 78–90, 2018. View at: Publisher Site | Google Scholar
  52. L. Sørensen and M. Nielsen, “Ensemble support vector machine classification of dementia using structural MRI and mini-mental state examination,” Journal of Neuroscience Methods, vol. 302, pp. 66–74, 2018. View at: Publisher Site | Google Scholar
  53. V. Wasule and P. Sonar, “Classification of brain MRI using SVM and KNN classifier,” in Proceedings of the 2017 Third International Conference on Sensing, Signal Processing and Security (ICSSS), pp. 218–223, Chennai, India, May 2017. View at: Publisher Site | Google Scholar
  54. S. Lahmiri, D. A. Dawson, and A. Shmuel, “Performance of machine learning methods in diagnosing Parkinson’s disease based on dysphonia measures,” Biomedical Engineering Letters, vol. 8, no. 1, pp. 29–39, 2018. View at: Publisher Site | Google Scholar
  55. A. Abós, H. C. Baggio, B. Segura et al., “Discriminating cognitive status in Parkinson’s disease through functional connectomics and machine learning,” Scientific Reports, vol. 7, no. 1, Article ID 45347, 2017. View at: Publisher Site | Google Scholar
  56. Y. Yoo, “On predicting epileptic seizures from intracranial electroencephalography,” Biomedical Engineering Letters, vol. 7, no. 1, pp. 1–5, 2017. View at: Publisher Site | Google Scholar
  57. M. Montembeault, S. M. Brambati, F. Lamari et al., “Atrophy, metabolism and cognition in the posterior cortical atrophy spectrum based on Alzheimer’s disease cerebrospinal fluid biomarkers,” NeuroImage: Clinical, vol. 20, pp. 1018–1025, 2018. View at: Publisher Site | Google Scholar
  58. H.-P. Müller, F. Agosta, M. Gorges et al., “Cortico-efferent tract involvement in primary lateral sclerosis and amyotrophic lateral sclerosis: a two-centre tract of interest-based DTI analysis,” NeuroImage: Clinical, vol. 20, pp. 1062–1069, 2018. View at: Publisher Site | Google Scholar
  59. M. Lojowska, S. Ling, K. Roelofs, and E. J. Hermans, “Visuocortical changes during a freezing-like state in humans,” NeuroImage, vol. 179, pp. 313–325, 2018. View at: Publisher Site | Google Scholar
  60. N. Amoroso, M. La Rocca, A. Monaco, R. Bellotti, and S. Tangaro, “Complex networks reveal early MRI markers of Parkinson’s disease,” Medical Image Analysis, vol. 48, pp. 12–24, 2018. View at: Publisher Site | Google Scholar
  61. J. E. Han, N. Boachie, I. Garcia-Garcia, A. Michaud, and A. Dagher, “Neural correlates of dietary self-control in healthy adults: a meta-analysis of functional brain imaging studies,” Physiology & Behavior, vol. 192, pp. 98–108, 2018. View at: Publisher Site | Google Scholar
  62. C. Gallea, P. Herath, V. Voon et al., “Loss of inhibition in sensorimotor networks in focal hand dystonia,” NeuroImage: Clinical, vol. 17, pp. 90–97, 2018. View at: Publisher Site | Google Scholar
  63. H. Rubin-Falcone, F. Zanderigo, B. Thapa-Chhetry et al., “Pattern recognition of magnetic resonance imaging-based gray matter volume measurements classifies bipolar disorder and major depressive disorder,” Journal of Affective Disorders, vol. 227, pp. 498–505, 2018. View at: Publisher Site | Google Scholar
  64. X. Gao, W. Zhang, L. Yao et al., “Association between structural and functional brain alterations in drug-free patients with schizophrenia: a multimodal meta-analysis,” Journal of Psychiatry & Neuroscience, vol. 43, no. 2, pp. 131–142, 2018. View at: Publisher Site | Google Scholar
  65. J. Qiao, A. Li, C. Cao et al., “Aberrant functional network connectivity as a biomarker of generalized anxiety disorder,” Frontiers in Human Neuroscience, vol. 11, no. 626, pp. 1–10, 2017. View at: Publisher Site | Google Scholar
  66. Q. Xiang, Y. Wang, J. Zhang et al., “Progressive brain changes in the early stage of schizophrenia: a combined structural MRI and DTI study,” Neuropsychiatry (London), vol. 8, no. 2, pp. 523–532, 2018. View at: Publisher Site | Google Scholar
  67. C. B. McNabb, R. J. Tait, M. E. McIlwain et al., “Functional network dysconnectivity as a biomarker of treatment resistance in schizophrenia,” Schizophrenia Research, vol. 195, pp. 160–167, 2018. View at: Publisher Site | Google Scholar
  68. J. Sui, G. D. Pearlson, Y. Du et al., “In search of multimodal neuroimaging biomarkers of cognitive deficits in schizophrenia,” Biological Psychiatry, vol. 78, no. 11, pp. 794–804, 2015. View at: Publisher Site | Google Scholar
  69. N. K. Focke, M. Yogarajah, M. R. Symms, O. Gruber, W. Paulus, and J. S. Duncan, “Automated MR image classification in temporal lobe epilepsy,” Neuroimage, vol. 59, no. 1, pp. 356–362, 2012. View at: Publisher Site | Google Scholar
  70. J. D. Rudie, J. B. Colby, and N. Salamon, “Machine learning classification of mesial temporal sclerosis in epilepsy patients,” Epilepsy Research, vol. 117, pp. 63–69, 2015. View at: Publisher Site | Google Scholar
  71. B. C. Bernhardt, S.-J. Hong, A. Bernasconi, and N. Bernasconi, “Magnetic resonance imaging pattern learning in temporal lobe epilepsy: classification and prognostics,” Annals of Neurology, vol. 77, no. 3, pp. 436–446, 2015. View at: Publisher Site | Google Scholar
  72. N. Memarian, S. Kim, S. Dewar, J. Engel, and R. J. Staba, “Multimodal data and machine learning for surgery outcome prediction in complicated cases of mesial temporal lobe epilepsy,” Computers in Biology and Medicine, vol. 64, pp. 67–78, 2015. View at: Publisher Site | Google Scholar
  73. C. Davatzikos, D. Shen, R. C. Gur et al., “Whole-brain morphometric study of schizophrenia revealing a spatially complex set of focal abnormalities,” Archives of General Psychiatry, vol. 62, no. 11, pp. 1218–1227, 2005. View at: Publisher Site | Google Scholar
  74. K. Yasuhiro, S. Michio, K. Ferath et al., “Multivariate voxel-based morphometry successfully differentiates schizophrenia patients from healthy controls,” Neuroimage, vol. 34, no. 1, pp. 235–242, 2007. View at: Publisher Site | Google Scholar
  75. D. Schwarz and T. Kasparek, “Brain morphometry of MR images for automated classification of first-episode schizophrenia,” Information Fusion, vol. 19, pp. 97–102, 2014. View at: Publisher Site | Google Scholar
  76. B. Dietsche, T. Kircher, and I. Falkenberg, “Structural brain changes in schizophrenia at different stages of the illness: a selective review of longitudinal magnetic resonance imaging studies,” Australian & New Zealand Journal of Psychiatry, vol. 51, no. 5, pp. 500–508, 2017. View at: Publisher Site | Google Scholar
  77. A. M. D. Moura, W. H. L. Pinaya, A. Gadelha et al., “Investigating brain structural patterns in first episode psychosis and schizophrenia using MRI and a machine learning approach,” Psychiatry Research Neuroimaging, vol. 275, pp. 14–20, 2018. View at: Publisher Site | Google Scholar
  78. C. Tas, H. Mogulkoc, G. Eryilmaz et al., “Discriminating schizophrenia and schizo-obsessive disorder: a structural MRI study combining VBM and machine learning methods,” Neural Computing and Applications, vol. 29, no. 2, pp. 377–387, 2018. View at: Publisher Site | Google Scholar
  79. U. K. Haukvik, C. B. Hartberg, and I. Agartz, “Schizophrenia—what does structural MRI show?” Tidsskrift for Den Norske Legeforening, vol. 133, no. 8, pp. 850–853, 2013. View at: Publisher Site | Google Scholar

Copyright © 2020 ZhiHong Chen et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


More related articles

 PDF Download Citation Citation
 Download other formatsMore
 Order printed copiesOrder
Views709
Downloads381
Citations

Related articles

We are committed to sharing findings related to COVID-19 as quickly as possible. We will be providing unlimited waivers of publication charges for accepted research articles as well as case reports and case series related to COVID-19. Review articles are excluded from this waiver policy. Sign up here as a reviewer to help fast-track new submissions.