Support vector machine learning-based fMRI data group analysis

To explore the multivariate nature of fMRI data and to consider the inter-subject brain response discrepancies, a multivariate and brain response model-free method is fundamentally required. Two such methods are presented in this paper by integrating a machine learning algorithm, the support vector machine (SVM), and the random effect model. Without any brain response modeling, SVM was used to extract a whole brain spatial discriminance map (SDM), representing the brain response difference between the contrasted experimental conditions. Population inference was then obtained through the random effect analysis (RFX) or permutation testing (PMU) on the individual subjects' SDMs. Applied to arterial spin labeling (ASL) perfusion fMRI data, SDM RFX yielded lower false-positive rates in the null hypothesis test and higher detection sensitivity for synthetic activations with varying cluster size and activation strengths, compared to the univariate general linear model (GLM)-based RFX. For a sensory-motor ASL fMRI study, both SDM RFX and SDM PMU yielded similar activation patterns to GLM RFX and GLM PMU, respectively, but with higher t values and cluster extensions at the same significance level. Capitalizing on the absence of temporal noise correlation in ASL data, this study also incorporated PMU in the individual-level GLM and SVM analyses accompanied by group-level analysis through RFX or group-level PMU. Providing inferences on the probability of being activated or deactivated at each voxel, these individual-level PMU-based group analysis methods can be used to threshold the analysis results of GLM RFX, SDM RFX or SDM PMU.     

Hyperspectral data classification using geostatistics and support vector machines

Hyperspectral imagery combined with spatial features holds promise for improved remote sensing classification. In this letter, we propose a method for classification of hyperspectral data based on the incorporation of spatial arrangement of pixel's values. We use the semivariogram to measure the spatial correlation which is then combined with spectral features within the stacked kernel support vector machine framework. The proposed method is compared with a classifier based on first-order statistics. The overall classification accuracy is tested for the AVIRIS Indian Pines benchmark dataset. Error matrices are used to estimate individual class accuracy. Statistical significance of the accuracy estimates is assessed based on the kappa coefficient and z-statistics at the 95% confidence level. Empirical results show that the proposed approach gives better performance than the method based on first-order statistics.     

Combining multivariate voxel selection and support vector machines for mapping and classification of fMRI spatial patterns

In functional brain mapping, pattern recognition methods allow detecting multivoxel patterns of brain activation which are informative with respect to a subject's perceptual or cognitive state. The sensitivity of these methods, however, is greatly reduced when the proportion of voxels that convey the discriminative information is small compared to the total number of measured voxels. To reduce this dimensionality problem, previous studies employed univariate voxel selection or region-of-interest-based strategies as a preceding step to the application of machine learning algorithms. Here we employ a strategy for classifying functional imaging data based on a multivariate feature selection algorithm, Recursive Feature Elimination (RFE) that uses the training algorithm (support vector machine) recursively to eliminate irrelevant voxels and estimate informative spatial patterns. Generalization performances on test data increases while features/voxels are pruned based on their discrimination ability. In this article we evaluate RFE in terms of sensitivity of discriminative maps (Receiver Operative Characteristic analysis) and generalization performances and compare it to previously used univariate voxel selection strategies based on activation and discrimination measures. Using simulated fMRI data, we show that the recursive approach is suitable for mapping discriminative patterns and that the combination of an initial univariate activation-based (F-test) reduction of voxels and multivariate recursive feature elimination produces the best results, especially when differences between conditions have a low contrast-to-noise ratio. Furthermore, we apply our method to high resolution (2 x 2 x 2 mm(3)) data from an auditory fMRI experiment in which subjects were stimulated with sounds from four different categories. With these real data, our recursive algorithm proves able to detect and accurately classify multivoxel spatial patterns, highlighting the role of the superior temporal gyrus in encoding the information of sound categories. In line with the simulation results, our method outperforms univariate statistical analysis and statistical learning without feature selection.     

A new classification model for a class imbalanced data set using genetic programming and support vector machines: case study for wilt disease classification

Class imbalanced data set is a state where each class of the given data set is not evenly distributed. When such case happens, most standard classifiers fail to recognize examples that belong to a minority class. Hence, several methods have been proposed to solve this problem such as resampling, modification on classifier optimization problem or introducing a new optimization task on top of the classifier. This work proposes a new optimization task based on genetic programming, built on top of support vector machine, in order to improve the classification rate for minority class without significant reduction on accuracy metric. The experimentation carried out on wilt disease data set shows the new classifier, support vector based on genetic programming machine, gives a more balanced accuracy between classes compared to various classification techniques in solving the imbalanced classification problem.     

A regularization algorithm for decoding perceptual temporal profiles from fMRI data

In several biomedical fields, researchers are faced with regression problems that can be stated as Statistical Learning problems. One example is given by decoding brain states from functional magnetic resonance imaging (fMRI) data. Recently, it has been shown that the general Statistical Learning problem can be restated as a linear inverse problem. Hence, new algorithms were proposed to solve this inverse problem in the context of Reproducing Kernel Hilbert Spaces. In this paper, we detail one iterative learning algorithm belonging to this class, called ν-method, and test its effectiveness in a between-subjects regression framework. Specifically, our goal was to predict the perceived pain intensity based on fMRI signals, during an experimental model of acute prolonged noxious stimulation. We found that, using a linear kernel, the psychophysical time profile was well reconstructed, while pain intensity was in some cases significantly over/underestimated. No substantial differences in terms of accuracy were found between the proposed approach and one of the state-of-the-art learning methods, the Support Vector Machines. Nonetheless, adopting the ν-method yielded a significant reduction in computational time, an advantage that became more evident when a relevant feature selection procedure was implemented. The ν-method can be easily extended and included in typical approaches for binary or multiple classification problems, and therefore it seems well-suited to build effective brain activity estimators.     

Power calculation for group fMRI studies accounting for arbitrary design and temporal autocorrelation

When planning most scientific studies, one of the first steps is to carry out a power analysis to define a design and sample size that will result in a well-powered study. There are limited resources for calculating power for group fMRI studies due to the complexity of the model. Previous approaches for group fMRI power calculation simplify the study design and/or the variance structure in order to make the calculation possible. These approaches limit the designs that can be studied and may result in inaccurate power calculations. We introduce a flexible power calculation model that makes fewer simplifying assumptions, leading to a more accurate power analysis that can be used on a wide variety of study designs. Our power calculation model can be used to obtain region of interest (ROI) summaries of the mean parameters and variance parameters, which can be use to increase understanding of the data as well as calculate power for a future study. Our example illustrates that minimizing cost to achieve 80% power is not as simple as finding the smallest sample size capable of achieving 80% power, since smaller sample sizes require each subject to be scanned longer.     

Optimal classification for the diagnosis of duchenne muscular dystrophy images using support vector machines

Background

This study aimed to investigate the optimal support vector machines (SVM)-based classifier of duchenne muscular dystrophy (DMD) magnetic resonance imaging (MRI) images.

Methods

T1-weighted (T1W) and T2-weighted (T2W) images of the 15 boys with DMD and 15 normal controls were obtained. Textural features of the images were extracted and wavelet decomposed, and then, principal features were selected. Scale transform was then performed for MRI images. Afterward, SVM-based classifiers of MRI images were analyzed based on the radical basis function and decomposition levels. The cost (C) parameter and kernel parameter \(\gamma \) were used for classification. Then, the optimal SVM-based classifier, expressed as \((C,\gamma \)), was identified by performance evaluation (sensitivity, specificity and accuracy).

Results

Eight of 12 textural features were selected as principal features (eigenvalues \(\lambda _{\mathrm{c}}\ge 1\)). The 16 SVM-based classifiers were obtained using combination of (C, \(\gamma \)), and those with lower C and \(\gamma \) values showed higher performances, especially classifier of \((C = 1,\gamma = 0.083)\,(p<0.05\)). The SVM-based classifiers of T1W images showed higher performance than T1W images at the same decomposition level. The T1W images in classifier of \((C = 1,\gamma = 0.083\)) at level 2 decomposition showed the highest performance of all, and its overall correct sensitivity, specificity, and accuracy reached 96.9, 97.3, and 97.1 %, respectively.

Conclusion

The T1W images in SVM-based classifier \((C =1, \gamma = 0.083)\) at level 2 decomposition showed the highest performance of all, demonstrating that it was the optimal classification for the diagnosis of DMD.

     

模糊聚类支持向量机在步态分类中的应用

背景：对于患有神经系统或骨骼肌肉系统疾病的患者,分析步态数据可以评定康复程度,制定治疗方案。如何有效地分类小样本步态数据成为重要的研究课题。目的：用改进的支持向量机算法对小样本步态数据进行分类,准确诊断疾病。方法：建立加入模糊C均值聚类的支持向量机算法,选用Gait Dynamics in Neuro-Degenerative Disease Data Base40～59岁年龄段的6组数据,共720个样本数据,采用左摆间隔和左支撑间隔两维参数对步态数据建模。数据归一化后,通过模糊C均值聚类对数据进行预处理;然后用支持向量机对数据进行分类。采用不同核函数的支持向量机算法验证分类能力。结果与结论：实验结果表明,利用改进的支持向量机算法,可以有效地对信号进行分类,有助于疾病的诊断和治疗方案的制定。     

模糊聚类支持向量机在步态分类中的应用

背景:对于患有神经系统或骨骼肌肉系统疾病的患者,分析步态数据可以评定康复程度,制定治疗方案。如何有效地分类小样本步态数据成为重要的研究课题。目的:用改进的支持向量机算法对小样本步态数据进行分类,准确诊断疾病。方法:建立加入模糊C均值聚类的支持向量机算法,选用Gait Dynamics in Neuro-Degenerative Disease Data Base40～59岁年龄段的6组数据,共720个样本数据,采用左摆间隔和左支撑间隔两维参数对步态数据建模。数据归一化后,通过模糊C均值聚类对数据进行预处理;然后用支持向量机对数据进行分类。采用不同核函数的支持向量机算法验证分类能力。结果与结论:实验结果表明,利用改进的支持向量机算法,可以有效地对信号进行分类,有助于疾病的诊断和治疗方案的制定。     

用于功能磁共振成像数据处理的时间簇分析法

作为一种新的数据处理方法,在大脑激活的位置和时间信息完全未知的情况下,时间簇分析(TCA)可以基于每个时间点达到最大值的像素的个数或灰度值得到大脑激活的时间信息.该方法具有算法容易实现,计算量小等优点,但其灵敏度低.目前对该方法的研究还停留在单层磁共振数据上,限制了其应用.     

Measuring temporal dynamics of functional networks using phase spectrum of fMRI data

We present a novel method to measure relative latencies between functionally connected regions using phase-delay of functional magnetic resonance imaging data. Derived from the phase component of coherency, this quantity estimates the linear delay between two time-series. In conjunction with coherence, derived from the magnitude component of coherency, phase-delay can be used to examine the temporal properties of functional networks. In this paper, we apply coherence and phase-delay methods to fMRI data in order to investigate dynamics of the motor network during task and rest periods. Using the supplementary motor area (SMA) as a reference region, we calculated relative latencies between the SMA and other regions within the motor network including the dorsal premotor cortex (PMd), primary motor cortex (M1), and posterior parietal cortex (PPC). During both the task and rest periods, we measured significant delays that were consistent across subjects. Specifically, we found significant delays between the SMA and the bilateral PMd, bilateral M1, and bilateral PPC during the task condition. During the rest condition, we found that the temporal dynamics of the network changed relative to the task period. No significant delays were measured between the SMA and the left PM and left M1; however, the right PM, right M1, and bilateral PPC were significantly delayed with respect to the SMA. Additionally, we observed significant map-wise differences in the dynamics of the network at task compared to the network at rest. These differences were observed in the interaction between the SMA and the left M1, left superior frontal gyrus, and left middle frontal gyrus. These temporal measurements are important in determining how regions within a network interact and provide valuable information about the sequence of cognitive processes within a network.     

Influence of fMRI data sampling on the temporal characterization of the hemodynamic response

Experimental and modeling studies were used to estimate the effect of different sampling rates (repetition times, TR) and different sampling positions on the estimates of the temporal properties of the hemodynamic response function (HRF) derived from fMRI studies. Data were acquired at a TR of 250 ms and then subjected to various degrees of undersampling. Using a gaussian fitting function it is demonstrated that the accuracy of HRF peak time determination decreases with lower sampling rate (higher TR). The decrease in accuracy amounts to about 50 ms per second of TR increase. In addition, temporal shifts of the HRF peak time are found when reducing the influence of the more variable descending part of HRF curve by using a temporal cut-off after HRF peak time. The shift scales with TR, amounts up to 100 ms for a TR of 1500 ms and a cut-off of 3-4 s and depends on the sampling position. The use of the full HRF function does not lead to a shift but increases the influence of potential confounding factors as large veins and poststimulus undershoot. Since both accuracy and potential shifts of HRF peak determination scale with TR, it is important that temporal fMRI studies are carried out with high sampling rates.     

Spatial and temporal reproducibility-based ranking of the independent components of BOLD fMRI data

Independent component analysis (ICA) decomposes fMRI data into spatially independent maps and their corresponding time courses. However, distinguishing the neurobiologically and biophysically reasonable components from those representing noise and artifacts is not trivial. We present a simple method for the ranking of independent components, by assessing the resemblance between components estimated from all the data, and components estimated from only the odd- (or even-) numbered time points. We show that the meaningful independent components of fMRI data resemble independent components estimated from downsampled data, and thus tend to be highly ranked by the method.     

Support vector machine classification and characterization of age-related reorganization of functional brain networks

Most of what is known about the reorganization of functional brain networks that accompanies normal aging is based on neuroimaging studies in which participants perform specific tasks. In these studies, reorganization is defined by the differences in task activation between young and old adults. However, task activation differences could be the result of differences in task performance, strategy, or motivation, and not necessarily reflect reorganization. Resting-state fMRI provides a method of investigating functional brain networks without such confounds. Here, a support vector machine (SVM) classifier was used in an attempt to differentiate older adults from younger adults based on their resting-state functional connectivity. In addition, the information used by the SVM was investigated to see what functional connections best differentiated younger adult brains from older adult brains. Three separate resting-state scans from 26 younger adults (18-35 yrs) and 26 older adults (55-85) were obtained from the International Consortium for Brain Mapping (ICBM) dataset made publically available in the 1000 Functional Connectomes project www.nitrc.org/projects/fcon_1000. 100 seed-regions from four functional networks with 5mm(3) radius were defined based on a recent study using machine learning classifiers on adolescent brains. Time-series for every seed-region were averaged and three matrices of z-transformed correlation coefficients were created for each subject corresponding to each individual's three resting-state scans. SVM was then applied using leave-one-out cross-validation. The SVM classifier was 84% accurate in classifying older and younger adult brains. The majority of the connections used by the classifier to distinguish subjects by age came from seed-regions belonging to the sensorimotor and cingulo-opercular networks. These results suggest that age-related decreases in positive correlations within the cingulo-opercular and default networks, and decreases in negative correlations between the default and sensorimotor networks, are the distinguishing characteristics of age-related reorganization.     

Multivariate model specification for fMRI data

We present a general method-denoted MoDef-to help specify (or define) the model used to analyze brain imaging data. This method is based on the use of the multivariate linear model on a training data set. We show that when the a priori knowledge about the expected brain response is not too precise, the method allows for the specification of a model that yields a better sensitivity in the statistical results. This obviously relies on the validity of the a priori information, in our case the representativity of the training set, an issue addressed using a cross-validation technique. We propose a fast implementation that allows the use of the method on large data sets as found with functional Magnetic Resonance Images. An example of application is given on an experimental fMRI data set that includes nine subjects who performed a mental computation task. Results show that the method increases the statistical sensitivity of fMRI analyses.     

SSPNet: An interpretable 3D-CNN for classification of schizophrenia using phase maps of resting-state complex-valued fMRI data

Convolutional neural networks (CNNs) have shown promising results in classifying individuals with mental disorders such as schizophrenia using resting-state fMRI data. However, complex-valued fMRI data is rarely used since additional phase data introduces high-level noise though it is potentially useful information for the context of classification. As such, we propose to use spatial source phase (SSP) maps derived from complex-valued fMRI data as the CNN input. The SSP maps are not only less noisy, but also more sensitive to spatial activation changes caused by mental disorders than magnitude maps. We build a 3D-CNN framework with two convolutional layers (named SSPNet) to fully explore the 3D structure and voxel-level relationships from the SSP maps. Two interpretability modules, consisting of saliency map generation and gradient-weighted class activation mapping (Grad-CAM), are incorporated into the well-trained SSPNet to provide additional information helpful for understanding the output. Experimental results from classifying schizophrenia patients (SZs) and healthy controls (HCs) show that the proposed SSPNet significantly improved accuracy and AUC compared to CNN using magnitude maps extracted from either magnitude-only (by 23.4 and 23.6% for DMN) or complex-valued fMRI data (by 10.6 and 5.8% for DMN). SSPNet captured more prominent HC-SZ differences in saliency maps, and Grad-CAM localized all contributing brain regions with opposite strengths for HCs and SZs within SSP maps. These results indicate the potential of SSPNet as a sensitive tool that may be useful for the development of brain-based biomarkers of mental disorders.     

功能磁共振成像时间序列分析中的RTSC方法

本文提出了二次修正的时间自相关(RTSC)方法,该方法引入了相关分析法中的刺激函数,与OTSC和MTSC方法相比,有效剔除了功能核磁共振成像中活动区域的抑制反应,更准确的检测出功能活动时脑内的激活区域.