M. Dewan, Y. Zhan, G. Hermosillo, B. Jian, X. Zhou
In the last decade, Brain PET Imaging has taken big strides in becoming an effective diagnostic tool for dementia and epilepsy disorders, particularly Alzheimer's. CT is often used to provide information for PET attenuation correction. However, for dementia patients, which often require multiple follow-ups, the elimination of CT is desirable to reduce the radiation dose. In this paper, we present a robust algorithm for PET attenuation correction without CT. The algorithm involves building a database of non-attenuation corrected (NAC) PET and CT pairs (model scans). Given a new patient's NAC PET, a learning-based algorithm is used to detect key landmarks, which are then used to select the most similar model scans. Deformable registration is then employed to warp the model CTs to the subject space, followed by a fusion step to obtain the virtual CT for attenuation correction. Besides comparing the normalized AC values with ground truth, we also use a diagnostic tool to evaluate the solution. In addition, a diagnostic evaluation is conducted by a trained nuclear medicine physician, all with promising results.
{"title":"Brain PET Attenuation Correction without CT: An Investigation","authors":"M. Dewan, Y. Zhan, G. Hermosillo, B. Jian, X. Zhou","doi":"10.1109/PRNI.2013.37","DOIUrl":"https://doi.org/10.1109/PRNI.2013.37","url":null,"abstract":"In the last decade, Brain PET Imaging has taken big strides in becoming an effective diagnostic tool for dementia and epilepsy disorders, particularly Alzheimer's. CT is often used to provide information for PET attenuation correction. However, for dementia patients, which often require multiple follow-ups, the elimination of CT is desirable to reduce the radiation dose. In this paper, we present a robust algorithm for PET attenuation correction without CT. The algorithm involves building a database of non-attenuation corrected (NAC) PET and CT pairs (model scans). Given a new patient's NAC PET, a learning-based algorithm is used to detect key landmarks, which are then used to select the most similar model scans. Deformable registration is then employed to warp the model CTs to the subject space, followed by a fusion step to obtain the virtual CT for attenuation correction. Besides comparing the normalized AC values with ground truth, we also use a diagnostic tool to evaluate the solution. In addition, a diagnostic evaluation is conducted by a trained nuclear medicine physician, all with promising results.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125821148","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}
H. E. Çetingül, Laura Dumont, M. Nadar, P. Thompson, G. Sapiro, C. Lenglet
We consider the problem of improving the accuracy and reliability of probabilistic white matter tractography methods by improving the built-in sampling scheme, which randomly draws, from a diffusion model such as the orientation distribution function (ODF), a direction of propagation. Existing methods employing inverse transform sampling require an ad hoc thresholding step to prevent the less likely directions from being sampled. We herein propose to perform importance sampling of spherical harmonics, which redistributes an input point set on the sphere to match the ODF using hierarchical sample warping. This produces a point set that is more concentrated around the modes, allowing the subsequent inverse transform sampling to generate orientations that are in better accordance with the local fiber configuration. Integrated into a Kalman filter-based framework, our approach is evaluated through experiments on synthetic, phantom, and real datasets.
{"title":"Importance Sampling Spherical Harmonics to Improve Probabilistic Tractography","authors":"H. E. Çetingül, Laura Dumont, M. Nadar, P. Thompson, G. Sapiro, C. Lenglet","doi":"10.1109/PRNI.2013.21","DOIUrl":"https://doi.org/10.1109/PRNI.2013.21","url":null,"abstract":"We consider the problem of improving the accuracy and reliability of probabilistic white matter tractography methods by improving the built-in sampling scheme, which randomly draws, from a diffusion model such as the orientation distribution function (ODF), a direction of propagation. Existing methods employing inverse transform sampling require an ad hoc thresholding step to prevent the less likely directions from being sampled. We herein propose to perform importance sampling of spherical harmonics, which redistributes an input point set on the sphere to match the ODF using hierarchical sample warping. This produces a point set that is more concentrated around the modes, allowing the subsequent inverse transform sampling to generate orientations that are in better accordance with the local fiber configuration. Integrated into a Kalman filter-based framework, our approach is evaluated through experiments on synthetic, phantom, and real datasets.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133692910","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}
M. Grosse-Wentrup, S. Harmeling, T. Zander, N. Hill, B. Scholkopf
We provide a simple method, based on volume conduction models, to quantify the neurophysiological plausibility of independent components (ICs) reconstructed from EEG/MEG data. We evaluate the method on EEG data recorded from 19 subjects and compare the results with two established procedures for judging the quality of ICs. We argue that our procedure provides a sound empirical basis for the inclusion or exclusion of ICs in the analysis of experimental data.
{"title":"How to Test the Quality of Reconstructed Sources in Independent Component Analysis (ICA) of EEG/MEG Data","authors":"M. Grosse-Wentrup, S. Harmeling, T. Zander, N. Hill, B. Scholkopf","doi":"10.1109/PRNI.2013.35","DOIUrl":"https://doi.org/10.1109/PRNI.2013.35","url":null,"abstract":"We provide a simple method, based on volume conduction models, to quantify the neurophysiological plausibility of independent components (ICs) reconstructed from EEG/MEG data. We evaluate the method on EEG data recorded from 19 subjects and compare the results with two established procedures for judging the quality of ICs. We argue that our procedure provides a sound empirical basis for the inclusion or exclusion of ICs in the analysis of experimental data.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132763844","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}
Developing multi-subject predictive models based on whole-brain neuroimage data for each subject is a major challenge due to the spatio-temporal nature of the variables and the massive amount of data relative to the number of subjects. We propose a novel multivariate machine learning model and algorithmic strategy for multi-subject regression or classification that uses regularization to directly account for the spatio-temporal nature of the data. Our method begins by fitting multi-subject models to each location separately (similar to univariate frameworks), and then aggregates information across nearby locations through regularization. We develop an optimization strategy so that our so called, Local-Aggregate Models, can be fit in a completely distributed manner over the locations which greatly reduces computational costs. Our models achieve better predictions with more interpretable results as demonstrated through a multi-subject EEG example.
{"title":"Local-Aggregate Modeling for Multi-subject Neuroimage Data via Distributed Optimization","authors":"Yue Hu, Genevera I. Allen","doi":"10.1109/PRNI.2013.60","DOIUrl":"https://doi.org/10.1109/PRNI.2013.60","url":null,"abstract":"Developing multi-subject predictive models based on whole-brain neuroimage data for each subject is a major challenge due to the spatio-temporal nature of the variables and the massive amount of data relative to the number of subjects. We propose a novel multivariate machine learning model and algorithmic strategy for multi-subject regression or classification that uses regularization to directly account for the spatio-temporal nature of the data. Our method begins by fitting multi-subject models to each location separately (similar to univariate frameworks), and then aggregates information across nearby locations through regularization. We develop an optimization strategy so that our so called, Local-Aggregate Models, can be fit in a completely distributed manner over the locations which greatly reduces computational costs. Our models achieve better predictions with more interpretable results as demonstrated through a multi-subject EEG example.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117351219","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}
N. Honnorat, H. Eavani, T. Satterthwaite, C. Davatzikos
fMRI is a powerful tool for assessing the functioning of the brain. The analysis of resting-state fMRI allows to describe the functional relationship between the cortical areas. Since most connectivity analysis methods suffer from the curse of dimensionality, the cortex needs to be first partitioned into regions of coherent activation patterns. Once the signals of these regions of interest have been extracted, estimating a sparse approximation of the inverse of their correlation matrix is a classical way to robustly describe their functional interactions. In this paper, we address both objectives with a novel parcellation method based on Markov Random Fields that favors the extraction of sparse networks of regions. Our method relies on state of the art rsfMRI models, naturally adapts the number of parcels to the data and is guaranteed to provide connected regions due to the use of shape priors. The second contribution of this paper resides in two novel sparsity enforcing potentials. Our approach is validated with a publicly available dataset.
{"title":"A Graph-Based Brain Parcellation Method Extracting Sparse Networks","authors":"N. Honnorat, H. Eavani, T. Satterthwaite, C. Davatzikos","doi":"10.1109/PRNI.2013.48","DOIUrl":"https://doi.org/10.1109/PRNI.2013.48","url":null,"abstract":"fMRI is a powerful tool for assessing the functioning of the brain. The analysis of resting-state fMRI allows to describe the functional relationship between the cortical areas. Since most connectivity analysis methods suffer from the curse of dimensionality, the cortex needs to be first partitioned into regions of coherent activation patterns. Once the signals of these regions of interest have been extracted, estimating a sparse approximation of the inverse of their correlation matrix is a classical way to robustly describe their functional interactions. In this paper, we address both objectives with a novel parcellation method based on Markov Random Fields that favors the extraction of sparse networks of regions. Our method relies on state of the art rsfMRI models, naturally adapts the number of parcels to the data and is guaranteed to provide connected regions due to the use of shape priors. The second contribution of this paper resides in two novel sparsity enforcing potentials. Our approach is validated with a publicly available dataset.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123453335","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}
Inferring functional connectivity, or statistical dependencies between activity in different regions of the brain, is of great interest in the study of neurocognitive conditions. For example, studies [1]-[3] indicate that patterns in connectivity might yield potential biomarkers for conditions such as Alzheimer's and autism. We model functional connectivity using Markov Networks, which use conditional dependence to determine when brain regions are directly connected. In this paper, we show that standard large-scale two-sample testing that compares graphs from distinct populations using subject level estimates of functional connectivity, fails to detect differences in functional connections. We propose a novel procedure to conduct two-sample inference via resampling and randomized edge selection to detect differential connections, with substantial improvement in statistical power and error control.
{"title":"Randomized Approach to Differential Inference in Multi-subject Functional Connectivity","authors":"Manjari Narayan, Genevera I. Allen","doi":"10.1109/PRNI.2013.29","DOIUrl":"https://doi.org/10.1109/PRNI.2013.29","url":null,"abstract":"Inferring functional connectivity, or statistical dependencies between activity in different regions of the brain, is of great interest in the study of neurocognitive conditions. For example, studies [1]-[3] indicate that patterns in connectivity might yield potential biomarkers for conditions such as Alzheimer's and autism. We model functional connectivity using Markov Networks, which use conditional dependence to determine when brain regions are directly connected. In this paper, we show that standard large-scale two-sample testing that compares graphs from distinct populations using subject level estimates of functional connectivity, fails to detect differences in functional connections. We propose a novel procedure to conduct two-sample inference via resampling and randomized edge selection to detect differential connections, with substantial improvement in statistical power and error control.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128984659","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}
The hierarchical organization of brain networks can be captured by clustering time series using multiple numbers of clusters, or scales, in resting-state functional magnetic resonance imaging. However, the systematic examination of all scales is a tedious task. Here, I propose a method to select a limited number of scales that are representative of the full hierarchy. A bootstrap analysis is first performed to estimate stability matrices, which quantify the reliability of the clustering for every pair of brain regions, over a grid of possible scales. A subset of scales is then selected to approximate linearly all stability matrices with a specified level of accuracy. On real data, the method was found to select a relatively small (~7) number of scales to explain 95% of the energy of 73 scales ranging from 2 to 1100 clusters. The number of selected scales was very consistent across 43 subjects, and the actual scales also showed some good level of agreement. This approach thus provides a principled approach to mine hierarchical brain networks, in the form of a few scales amenable to detailed examination.
{"title":"Mining the Hierarchy of Resting-State Brain Networks: Selection of Representative Clusters in a Multiscale Structure","authors":"Pierre Bellec","doi":"10.1109/PRNI.2013.23","DOIUrl":"https://doi.org/10.1109/PRNI.2013.23","url":null,"abstract":"The hierarchical organization of brain networks can be captured by clustering time series using multiple numbers of clusters, or scales, in resting-state functional magnetic resonance imaging. However, the systematic examination of all scales is a tedious task. Here, I propose a method to select a limited number of scales that are representative of the full hierarchy. A bootstrap analysis is first performed to estimate stability matrices, which quantify the reliability of the clustering for every pair of brain regions, over a grid of possible scales. A subset of scales is then selected to approximate linearly all stability matrices with a specified level of accuracy. On real data, the method was found to select a relatively small (~7) number of scales to explain 95% of the energy of 73 scales ranging from 2 to 1100 clusters. The number of selected scales was very consistent across 43 subjects, and the actual scales also showed some good level of agreement. This approach thus provides a principled approach to mine hierarchical brain networks, in the form of a few scales amenable to detailed examination.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124995276","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}
Oluwasanmi Koyejo, Priyank Patel, Joydeep Ghosh, R. Poldrack
We present an experimental study of topic models applied to the analysis of functional magnetic resonance images. This study is motivated by the hypothesis that experimental task contrast images share a common set of mental concepts. We represent the images as documents and the mental concepts as topics, and evaluate the effectiveness of unsupervised topic models for the recovery of the task to mental concept mapping, We also evaluate supervised topic models that explicitly incorporate the experimental task labels. Comparing the quality of the recovered topic assignments to known mental concepts, we find that the supervised models are more effective than unsupervised approaches. The quantitative performance results are supported by a visualization of the recovered topic assignment probabilities. Our results motivate the use of supervised topic models for analyzing cognitive function with fMRI.
{"title":"Learning Predictive Cognitive Structure from fMRI Using Supervised Topic Models","authors":"Oluwasanmi Koyejo, Priyank Patel, Joydeep Ghosh, R. Poldrack","doi":"10.1109/PRNI.2013.12","DOIUrl":"https://doi.org/10.1109/PRNI.2013.12","url":null,"abstract":"We present an experimental study of topic models applied to the analysis of functional magnetic resonance images. This study is motivated by the hypothesis that experimental task contrast images share a common set of mental concepts. We represent the images as documents and the mental concepts as topics, and evaluate the effectiveness of unsupervised topic models for the recovery of the task to mental concept mapping, We also evaluate supervised topic models that explicitly incorporate the experimental task labels. Comparing the quality of the recovered topic assignments to known mental concepts, we find that the supervised models are more effective than unsupervised approaches. The quantitative performance results are supported by a visualization of the recovered topic assignment probabilities. Our results motivate the use of supervised topic models for analyzing cognitive function with fMRI.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122540196","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}
Network organization is fundamental to the human brain and alterations of this organization by brain states and neurological diseases is an active field of research. Many studies investigate functional networks by considering temporal correlations between the fMRI signal of distinct brain regions over long periods of time. Here, we propose to use the higher-order singular value decomposition (HOSVD), a tensor decomposition, to extract whole-brain network signatures from group-level dynamic functional connectivity data. HOSVD is a data-driven multivariate method that fits the data to a 3-way model, i.e., connectivity x time x subjects. We apply the proposed method to fMRI data with alternating epochs of resting and watching of movie excerpts, where we captured dynamic functional connectivity by sliding window correlations. By regressing the connectivity maps' time courses with the experimental paradigm, we find a characteristic connectivity pattern for the difference between the brain states. Using leave-one-subject-out cross-validation, we then show that the combination of connectivity patterns generalizes to unseen subjects as it predicts the paradigm. The proposed technique can be used as feature extraction for connectivity-based decoding and holds promise for the study of dynamic brain networks.
{"title":"Identifying Network Correlates of Brain States Using Tensor Decompositions of Whole-Brain Dynamic Functional Connectivity","authors":"Nora Leonardi, D. Ville","doi":"10.1109/PRNI.2013.28","DOIUrl":"https://doi.org/10.1109/PRNI.2013.28","url":null,"abstract":"Network organization is fundamental to the human brain and alterations of this organization by brain states and neurological diseases is an active field of research. Many studies investigate functional networks by considering temporal correlations between the fMRI signal of distinct brain regions over long periods of time. Here, we propose to use the higher-order singular value decomposition (HOSVD), a tensor decomposition, to extract whole-brain network signatures from group-level dynamic functional connectivity data. HOSVD is a data-driven multivariate method that fits the data to a 3-way model, i.e., connectivity x time x subjects. We apply the proposed method to fMRI data with alternating epochs of resting and watching of movie excerpts, where we captured dynamic functional connectivity by sliding window correlations. By regressing the connectivity maps' time courses with the experimental paradigm, we find a characteristic connectivity pattern for the difference between the brain states. Using leave-one-subject-out cross-validation, we then show that the combination of connectivity patterns generalizes to unseen subjects as it predicts the paradigm. The proposed technique can be used as feature extraction for connectivity-based decoding and holds promise for the study of dynamic brain networks.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133748254","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}
J. V. Schependom, M. D'hooge, Krista Cleynhens, M. D'hooghe, J. Keyser, G. Nagels
Cognitive impairment affects half of the multiple sclerosis (MS) patient population and is an important factor of quality of life. Cognitive impairment is, however, difficult to detect. Apart from the traditional features used in P300 experiments (e.g. amplitude and latency at different electrodes), we want to investigate the value of network-features on the classification of MS patients as cognitively intact or impaired. We included 305 MS patients, recruited at the National MS Center Melsbroek (Belgium). About half of them was denoted cognitively impaired (143). We divided this patient group in a training set (on which we used 10-fold cross validation) and an independent test set. Results are reported on this last group to increase the generalizability. We found the correlations linking electrodes from one hemisphere with the other significantly different between the two groups MS patients. Especially in the parietal region this difference was very significant (1.5E-12). Using a simple cutoff on this variable, lead to a Percentage Correctly Classified (PCC) of 0.70 and an Area Under Curve (AUC) of the Receiver Operator Curve (ROC) of 0.76. The network parameters that were calculated showed a comparable result for the total number of edges included in the network. Combining these features in a logistic regression model, artificial neural networks or Naive Bayes resulted in a PCC's of 0.68-0.70. These results support the recent suggestion that cognitive dysfunction in MS is caused by a disconnection mechanism in the cerebellum. We have obtained these results applying graph theoretical analyses on EEG data instead of the more common fMRI-analyses. The classification accuracy obtained is, however, not yet sufficient for application in clinical practice.
{"title":"Detection of Cognitive Impairment in MS Based on an EEG P300 Paradigm","authors":"J. V. Schependom, M. D'hooge, Krista Cleynhens, M. D'hooghe, J. Keyser, G. Nagels","doi":"10.1109/PRNI.2013.38","DOIUrl":"https://doi.org/10.1109/PRNI.2013.38","url":null,"abstract":"Cognitive impairment affects half of the multiple sclerosis (MS) patient population and is an important factor of quality of life. Cognitive impairment is, however, difficult to detect. Apart from the traditional features used in P300 experiments (e.g. amplitude and latency at different electrodes), we want to investigate the value of network-features on the classification of MS patients as cognitively intact or impaired. We included 305 MS patients, recruited at the National MS Center Melsbroek (Belgium). About half of them was denoted cognitively impaired (143). We divided this patient group in a training set (on which we used 10-fold cross validation) and an independent test set. Results are reported on this last group to increase the generalizability. We found the correlations linking electrodes from one hemisphere with the other significantly different between the two groups MS patients. Especially in the parietal region this difference was very significant (1.5E-12). Using a simple cutoff on this variable, lead to a Percentage Correctly Classified (PCC) of 0.70 and an Area Under Curve (AUC) of the Receiver Operator Curve (ROC) of 0.76. The network parameters that were calculated showed a comparable result for the total number of edges included in the network. Combining these features in a logistic regression model, artificial neural networks or Naive Bayes resulted in a PCC's of 0.68-0.70. These results support the recent suggestion that cognitive dysfunction in MS is caused by a disconnection mechanism in the cerebellum. We have obtained these results applying graph theoretical analyses on EEG data instead of the more common fMRI-analyses. The classification accuracy obtained is, however, not yet sufficient for application in clinical practice.","PeriodicalId":144007,"journal":{"name":"2013 International Workshop on Pattern Recognition in Neuroimaging","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115906287","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：