Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858543
J. Schrouff, B. Foster, Vinitha Rangarajan, C. Phillips, J. Mourão-Miranda, Joseph Parvizi
Recently machine learning models have been applied to neuroimaging data, which allow predictions about a variable of interest based on the pattern of activation or anatomy over a set of voxels. These pattern recognition based methods present clear benefits over classical (univariate) techniques, by providing predictions for unseen data, as well as the weights of each feature in the model. Machine learning methods have been applied to a range of data, from MRI to EEG. However, these multivariate techniques have scarcely been applied to electrocorticography (ECoG) data to investigate cognitive neuroscience questions. In this work, we used previously published ECoG data from 8 subjects to show that machine learning techniques can complement univariate techniques and be more sensitive to certain effects.
{"title":"Decoding memory processing from electro-corticography in human posteromedial cortex","authors":"J. Schrouff, B. Foster, Vinitha Rangarajan, C. Phillips, J. Mourão-Miranda, Joseph Parvizi","doi":"10.1109/PRNI.2014.6858543","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858543","url":null,"abstract":"Recently machine learning models have been applied to neuroimaging data, which allow predictions about a variable of interest based on the pattern of activation or anatomy over a set of voxels. These pattern recognition based methods present clear benefits over classical (univariate) techniques, by providing predictions for unseen data, as well as the weights of each feature in the model. Machine learning methods have been applied to a range of data, from MRI to EEG. However, these multivariate techniques have scarcely been applied to electrocorticography (ECoG) data to investigate cognitive neuroscience questions. In this work, we used previously published ECoG data from 8 subjects to show that machine learning techniques can complement univariate techniques and be more sensitive to certain effects.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"12 5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128596992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858530
Jeremy R. Manning, R. Ranganath, Waitsang Keung, N. Turk-Browne, J. Cohen, K. Norman, D. Blei
Recent work has revealed that cognitive processes are often reflected in patterns of functional connectivity throughout the brain (for review see [16]). However, examining functional connectivity patterns using traditional methods carries a substantial computational burden (of computing time and memory). Here we present a technique, termed Hierarchical topographic factor analysis, for efficiently discovering brain networks in large multi-subject neuroimaging datasets.
{"title":"Hierarchical topographic factor analysis","authors":"Jeremy R. Manning, R. Ranganath, Waitsang Keung, N. Turk-Browne, J. Cohen, K. Norman, D. Blei","doi":"10.1109/PRNI.2014.6858530","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858530","url":null,"abstract":"Recent work has revealed that cognitive processes are often reflected in patterns of functional connectivity throughout the brain (for review see [16]). However, examining functional connectivity patterns using traditional methods carries a substantial computational burden (of computing time and memory). Here we present a technique, termed Hierarchical topographic factor analysis, for efficiently discovering brain networks in large multi-subject neuroimaging datasets.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116714731","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858544
F. Segovia, C. Phillips
Positron Emission Tomography (PET) is a noninvasive medical imaging modality that provides information about physiological processes. Due to its ability to measure the brain metabolism, it is widely used to assist the diagnosis of neurodegenerative disorders such as Alzheimer's disease (AD) of Parkinsonism. In order to avoid the subjectivity inherent to the visual exploration of the images, several computer systems to analyze PET data were developed during the last years. However, dealing with the huge amount of information provided by PET imaging is still a challenge. In this work we present a novel methodology to analyze PET data that improves the automatic differentiation between controls and AD patients. First the images are divided into small regions or parcels, defined either anatomically, geometrically or randomly. Secondly, the accuray of each single region is estimated using a Support Vector Machine (SVM) classifier and a cross-validation approach. Finally, all the regions are assessed together using multiple kernel SVM with a kernel per region. The classifier is built so that the most discriminative regions have more weight in the final decision. This method was evaluated using a PET dataset that contained images from healthy controls and AD patients. The classification results obtained with the proposed approach outperformed two recently reported computer systems based on Principal Component Analysis and Independent Component Analysis.
{"title":"PET imaging analysis using a parcelation approach and multiple kernel classification","authors":"F. Segovia, C. Phillips","doi":"10.1109/PRNI.2014.6858544","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858544","url":null,"abstract":"Positron Emission Tomography (PET) is a noninvasive medical imaging modality that provides information about physiological processes. Due to its ability to measure the brain metabolism, it is widely used to assist the diagnosis of neurodegenerative disorders such as Alzheimer's disease (AD) of Parkinsonism. In order to avoid the subjectivity inherent to the visual exploration of the images, several computer systems to analyze PET data were developed during the last years. However, dealing with the huge amount of information provided by PET imaging is still a challenge. In this work we present a novel methodology to analyze PET data that improves the automatic differentiation between controls and AD patients. First the images are divided into small regions or parcels, defined either anatomically, geometrically or randomly. Secondly, the accuray of each single region is estimated using a Support Vector Machine (SVM) classifier and a cross-validation approach. Finally, all the regions are assessed together using multiple kernel SVM with a kernel per region. The classifier is built so that the most discriminative regions have more weight in the final decision. This method was evaluated using a PET dataset that contained images from healthy controls and AD patients. The classification results obtained with the proposed approach outperformed two recently reported computer systems based on Principal Component Analysis and Independent Component Analysis.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127018457","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858537
E. Olivetti, S. M. Kia, P. Avesani
Traditional approaches to create sensor-level maps from magnetoencephalographic (MEG) data rely on mass-univariate methods. In order to overcome some limitations of univariate approaches, multivariate approaches have been widely investigated, mostly based on the paradigm of classification. Recently a multivariate two-sample test called kernel two-sample test (KTST) has been proposed as an alternative to classification-based methods. Unfortunately the KTST lacks methods for neuroscientific interpretation of its result, e.g. in terms of sensor-level maps. In this work, we address this issue and we propose a cluster-based permutation kernel two-sample test (CBPKTST) to create sensor-level maps. Moreover we propose a procedure that massively reduces the computation so that maps can be produced in minutes. We report preliminary experiments on MEG data in which we show that the proposed procedure has much greater sensitivity than the traditional cluster-based permutation t-test.
{"title":"Sensor-level maps with the kernel two-sample test","authors":"E. Olivetti, S. M. Kia, P. Avesani","doi":"10.1109/PRNI.2014.6858537","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858537","url":null,"abstract":"Traditional approaches to create sensor-level maps from magnetoencephalographic (MEG) data rely on mass-univariate methods. In order to overcome some limitations of univariate approaches, multivariate approaches have been widely investigated, mostly based on the paradigm of classification. Recently a multivariate two-sample test called kernel two-sample test (KTST) has been proposed as an alternative to classification-based methods. Unfortunately the KTST lacks methods for neuroscientific interpretation of its result, e.g. in terms of sensor-level maps. In this work, we address this issue and we propose a cluster-based permutation kernel two-sample test (CBPKTST) to create sensor-level maps. Moreover we propose a procedure that massively reduces the computation so that maps can be produced in minutes. We report preliminary experiments on MEG data in which we show that the proposed procedure has much greater sensitivity than the traditional cluster-based permutation t-test.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130193359","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858521
Sofie Therese Hansen, L. K. Hansen
State of the art performance of 3D EEG imaging is based on reconstruction using spatial basis function repre-sentations. In this work we augment the Variational Garrote (VG) approach for sparse approximation to incorporate spatial basis functions. As VG handles the bias variance trade-off with cross-validation this approach is more automated than competing approaches such as Multiple Sparse Priors (Friston et al., 2008) or Champagne (Wipf et al., 2010) that require manual selection of noise level and auxiliary signal free data, respectively. Finally, we propose an unbiased estimator of the reproducibility of the reconstructed activation time course based on a split-half resampling protocol.
{"title":"EEG source reconstruction using sparse basis function representations","authors":"Sofie Therese Hansen, L. K. Hansen","doi":"10.1109/PRNI.2014.6858521","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858521","url":null,"abstract":"State of the art performance of 3D EEG imaging is based on reconstruction using spatial basis function repre-sentations. In this work we augment the Variational Garrote (VG) approach for sparse approximation to incorporate spatial basis functions. As VG handles the bias variance trade-off with cross-validation this approach is more automated than competing approaches such as Multiple Sparse Priors (Friston et al., 2008) or Champagne (Wipf et al., 2010) that require manual selection of noise level and auxiliary signal free data, respectively. Finally, we propose an unbiased estimator of the reproducibility of the reconstructed activation time course based on a split-half resampling protocol.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"69 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128890518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858529
G. Lohmann, Johannes Stelzer, V. Zuber, T. Buschmann, M. Erb, K. Scheffler
Traditionally fMRI data analysis aims at identifying brain areas in which the amplitude of the BOLD signal responds to experimental stimulations. However, since the brain acts as a network, we would expect differential effects on network topology. Therefore, the target of statistical inference should not only be individual voxels or brain areas but rather network connections. Here we introduce a new approach to correlation-based statistics in fMRI. At the heart of our approach is the concept of correlation bundles as a functional analogy to anatomical fibre bundles. Statistical tests are applied to these bundles using large-scale inference methods such as FDR. We call this approach correlation bundle statistics (CBS). In contrast to previous correlation-based approaches to fMRI statistics, CBS does not require a presegmentation or smoothing of the data so that anatomical specificity is preserved. The result of a CBS analysis is not a set of voxels or brain regions but rather a set of correlation bundles that are found to be significantly affected by some experimental manipulation.
{"title":"Correlation bundle statistics in fMRI data","authors":"G. Lohmann, Johannes Stelzer, V. Zuber, T. Buschmann, M. Erb, K. Scheffler","doi":"10.1109/PRNI.2014.6858529","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858529","url":null,"abstract":"Traditionally fMRI data analysis aims at identifying brain areas in which the amplitude of the BOLD signal responds to experimental stimulations. However, since the brain acts as a network, we would expect differential effects on network topology. Therefore, the target of statistical inference should not only be individual voxels or brain areas but rather network connections. Here we introduce a new approach to correlation-based statistics in fMRI. At the heart of our approach is the concept of correlation bundles as a functional analogy to anatomical fibre bundles. Statistical tests are applied to these bundles using large-scale inference methods such as FDR. We call this approach correlation bundle statistics (CBS). In contrast to previous correlation-based approaches to fMRI statistics, CBS does not require a presegmentation or smoothing of the data so that anatomical specificity is preserved. The result of a CBS analysis is not a set of voxels or brain regions but rather a set of correlation bundles that are found to be significantly affected by some experimental manipulation.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125444866","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858541
J. V. Schependom, J. Gielen, J. Laton, M. D'hooghe, J. Keyser, G. Nagels
Cognitive impairment affects half of the multiple sclerosis (MS) patient population, is difficult to detect and requires extensive neuropsychological testing. We analyzed data obtained in a P300 experiment. The P300 is a large positive wave following an unexpected stimulus and is mainly related to attention, a domain frequently impaired in MS. Apart from the traditional features used in P300 experiments we want to investigate the value of different connectivity measures on the classification of MS patients as cognitively intact or impaired. We included 331 MS patients, recruited at the National MS Center Melsbroek (Belgium). About one third was denoted cognitively impaired (104). We divided our patient cohort in a training set (on which we used 10-fold crossvalidation) to optimize the (hyper)parameters of the SVM and an independent test set. Results are reported on this last group to increase the generalizability. In recent years many effort has been devoted to devising connectivity metrics for EEG and MEG data. The most commonly applied metrics are correlation and coherence. However, other metrics have been constructed like the Phase Lag Index (PLI) and the imaginary part of coherency (ImagCoh). Using traditional P300 features, we obtained an accuracy of 68 %. Several connectivity metrics returned similar results, especially the more traditional ones like correlation, correlation in the frequency domain and coherence (delta). The obtained accuracies were, however, only a minor improvement on the accuracy obtained using the traditional P300 features. These results support the recent suggestion that cognitive dysfunction in MS might be caused by cerebral disconnection. We have obtained these results applying graph theoretical analyses on EEG data instead of the more common fMRI network analyses. Although the classification accuracy denotes an important link to cognitive status, it is not sufficient for application in clinical practice.
{"title":"SVM aided detection of cognitive impairment in MS","authors":"J. V. Schependom, J. Gielen, J. Laton, M. D'hooghe, J. Keyser, G. Nagels","doi":"10.1109/PRNI.2014.6858541","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858541","url":null,"abstract":"Cognitive impairment affects half of the multiple sclerosis (MS) patient population, is difficult to detect and requires extensive neuropsychological testing. We analyzed data obtained in a P300 experiment. The P300 is a large positive wave following an unexpected stimulus and is mainly related to attention, a domain frequently impaired in MS. Apart from the traditional features used in P300 experiments we want to investigate the value of different connectivity measures on the classification of MS patients as cognitively intact or impaired. We included 331 MS patients, recruited at the National MS Center Melsbroek (Belgium). About one third was denoted cognitively impaired (104). We divided our patient cohort in a training set (on which we used 10-fold crossvalidation) to optimize the (hyper)parameters of the SVM and an independent test set. Results are reported on this last group to increase the generalizability. In recent years many effort has been devoted to devising connectivity metrics for EEG and MEG data. The most commonly applied metrics are correlation and coherence. However, other metrics have been constructed like the Phase Lag Index (PLI) and the imaginary part of coherency (ImagCoh). Using traditional P300 features, we obtained an accuracy of 68 %. Several connectivity metrics returned similar results, especially the more traditional ones like correlation, correlation in the frequency domain and coherence (delta). The obtained accuracies were, however, only a minor improvement on the accuracy obtained using the traditional P300 features. These results support the recent suggestion that cognitive dysfunction in MS might be caused by cerebral disconnection. We have obtained these results applying graph theoretical analyses on EEG data instead of the more common fMRI network analyses. Although the classification accuracy denotes an important link to cognitive status, it is not sufficient for application in clinical practice.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124480392","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858523
J. Höhne, B. Blankertz, K. Müller, Daniel Bartz
Linear discriminant analysis (LDA) is the most commonly used classification method for single trial data in a brain-computer interface (BCI) framework. The popularity of LDA arises from its robustness, simplicity and high accuracy. However, the standard LDA approach is not capable to exploit sublabel information (such as stimulus identity), which is accessible in data from event related potentials (ERPs): it assumes that the evoked potentials are independent of the stimulus identity and dependent only on the users' attentional state. We question this assumption and investigate several methods which extract subclass-specific features from ERP data. Moreover, we propose a novel classification approach which exploits subclass-specific features using mean shrinkage. Based on a reanalysis of two BCI data sets, we show that our novel approach outperforms the standard LDA approach, while being computationally highly efficient.
{"title":"Mean shrinkage improves the classification of ERP signals by exploiting additional label information","authors":"J. Höhne, B. Blankertz, K. Müller, Daniel Bartz","doi":"10.1109/PRNI.2014.6858523","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858523","url":null,"abstract":"Linear discriminant analysis (LDA) is the most commonly used classification method for single trial data in a brain-computer interface (BCI) framework. The popularity of LDA arises from its robustness, simplicity and high accuracy. However, the standard LDA approach is not capable to exploit sublabel information (such as stimulus identity), which is accessible in data from event related potentials (ERPs): it assumes that the evoked potentials are independent of the stimulus identity and dependent only on the users' attentional state. We question this assumption and investigate several methods which extract subclass-specific features from ERP data. Moreover, we propose a novel classification approach which exploits subclass-specific features using mean shrinkage. Based on a reanalysis of two BCI data sets, we show that our novel approach outperforms the standard LDA approach, while being computationally highly efficient.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123971465","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858514
Sven Dähne, V. Nikulin, D. Ramírez, P. Schreier, K. Müller, S. Haufe
Amplitude-to-amplitude interactions between neural oscillations are of a special interest as they show how the strength of spatial synchronization in different neuronal populations relates to each other during a given task. While, previously, amplitude-to-amplitude correlations were studied primarily on the sensor level, we present a source separation approach using spatial filters which maximize the correlation between the envelopes of brain oscillations recorded with electro-/magnetencephalography (EEG/MEG) or intracranial multichannel recordings. Our approach, which is called canonical source power correlation analysis (cSPoC), is thereby capable of extracting genuine brain oscillations solely based on their assumed coupling behavior even when the signal-to-noise ratio of the signals is low.
{"title":"Optimizing spatial filters for the extraction of envelope-coupled neural oscillations","authors":"Sven Dähne, V. Nikulin, D. Ramírez, P. Schreier, K. Müller, S. Haufe","doi":"10.1109/PRNI.2014.6858514","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858514","url":null,"abstract":"Amplitude-to-amplitude interactions between neural oscillations are of a special interest as they show how the strength of spatial synchronization in different neuronal populations relates to each other during a given task. While, previously, amplitude-to-amplitude correlations were studied primarily on the sensor level, we present a source separation approach using spatial filters which maximize the correlation between the envelopes of brain oscillations recorded with electro-/magnetencephalography (EEG/MEG) or intracranial multichannel recordings. Our approach, which is called canonical source power correlation analysis (cSPoC), is thereby capable of extracting genuine brain oscillations solely based on their assumed coupling behavior even when the signal-to-noise ratio of the signals is low.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"120 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121153205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-06-04DOI: 10.1109/PRNI.2014.6858553
Chuncheng Zhang, Zhengli Wang, Sutao Song, Xiaotong Wen, L. Yao, Zhi-ying Long
Feature selection (FS) plays an important role in improving the classification accuracy of multivariate classification techniques in the context of fMRI based decoding due to the “few samples and large features” of fMRI data. The multivariate FS methods are generally time-consuming although they displayed better performance than the univariate FS methods. In this study, we applied a fast sparse representation method based on Smoothed 10 (SLO) algorithm to select relevant features in fMRI data. The performance of Gaussian Naive Bayes (GNB) classifier using voxels selected by SLO and the univariate t-test methods were also compared. Results of both simulated and real fMRI experiments demonstrated that the SLO method largely improved the classification accuracy of GNB compared to the t-test method for all the noise levels.
{"title":"Fast voxel selection of fMRI data based on Smoothed 10 norm","authors":"Chuncheng Zhang, Zhengli Wang, Sutao Song, Xiaotong Wen, L. Yao, Zhi-ying Long","doi":"10.1109/PRNI.2014.6858553","DOIUrl":"https://doi.org/10.1109/PRNI.2014.6858553","url":null,"abstract":"Feature selection (FS) plays an important role in improving the classification accuracy of multivariate classification techniques in the context of fMRI based decoding due to the “few samples and large features” of fMRI data. The multivariate FS methods are generally time-consuming although they displayed better performance than the univariate FS methods. In this study, we applied a fast sparse representation method based on Smoothed 10 (SLO) algorithm to select relevant features in fMRI data. The performance of Gaussian Naive Bayes (GNB) classifier using voxels selected by SLO and the univariate t-test methods were also compared. Results of both simulated and real fMRI experiments demonstrated that the SLO method largely improved the classification accuracy of GNB compared to the t-test method for all the noise levels.","PeriodicalId":133286,"journal":{"name":"2014 International Workshop on Pattern Recognition in Neuroimaging","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130097737","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：