Cargando…

T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT

BACKGROUND: Impairments in cognitive functioning are a core feature of psychotic disorders and they have been associated with resting-state functional connectivity (rsFC) alterations in patients suffering from psychosis (Dauverman et al., 2014). Multivariate pattern analysis (MVPA) has proven to be...

Descripción completa

Detalles Bibliográficos
Autores principales: Weiske, Johanna, Ruef, Anne, Haas, Shalaila, Bonivento, Carolina, Koutsouleris, Nikolaos, Kambeitz-Ilankovic, Lana
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Oxford University Press 2018
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5887536/
http://dx.doi.org/10.1093/schbul/sby016.413
_version_ 1783312326011650048
author Weiske, Johanna
Ruef, Anne
Haas, Shalaila
Bonivento, Carolina
Koutsouleris, Nikolaos
Kambeitz-Ilankovic, Lana
author_facet Weiske, Johanna
Ruef, Anne
Haas, Shalaila
Bonivento, Carolina
Koutsouleris, Nikolaos
Kambeitz-Ilankovic, Lana
author_sort Weiske, Johanna
collection PubMed
description BACKGROUND: Impairments in cognitive functioning are a core feature of psychotic disorders and they have been associated with resting-state functional connectivity (rsFC) alterations in patients suffering from psychosis (Dauverman et al., 2014). Multivariate pattern analysis (MVPA) has proven to be a useful tool in the investigation of rsFC alteration in psychosis and in detecting subtle differences in multidimensional data sets (Kambeitz et al., 2015). In this study, we differentiated recent-onset psychosis patients (ROP) from healthy controls (HC) using a Support Vector Machine (SVM) classification based on rsFC. Furthermore, we investigated the relationship of the discriminative rsFC pattern to neurocognitive measures. METHODS: Resting-state fMRI and neurocognitive measures were obtained from 220 HC and 115 ROP across 7 sites of the PRONIA consortium. The rsFC matrix was estimated for each subject by calculating pairwise correlations between mean time series of 90 brain regions based on AAL parcellation. A L1-regularized L2-loss SVM was trained to classify ROP vs. HC based on rsFC in a repeated nested cross-validation. Decision scores for ROP were correlated with cognitive measures derived from the following neuropsychological tests: Rey Auditory Verbal Learning Task (RAVLT), Phonetic and Semantic Verbal Fluency, Diagnostic Analysis of Nonverbal Accuracy, Forward and Backward Digit Span, Self-ordered Pointing Task, and Salience Attribution Test. RESULTS: The classification algorithm was able to differentiate ROP and HC with a balanced accuracy (BAC) of 71.3% based on rsFC. The discriminative connectivity pattern included short-range connections between left putamen and left hippocampus, right putamen and right caudate nucleus, left superior frontal and right inferior orbitofrontal regions, as well as long-range connections between left and right occipital cortex and left cingulate gyrus, left supramarginal gyrus and right temporal pole. Two negative correlations between the SVM decision scores for ROP and measures of the RAVLT were significant (delayed recall: r=-0.3, Bonferroni –adjusted p<.04; recall after interference: r=-0.3, Bonferroni-adjusted p<.02). DISCUSSION: The classification performance was driven by a rsFC pattern including areas involved in memory processing, such as hippocampus and cingulate gyrus (Allen et al., 2007) as well as regions related to language processing, such as the supra marginal gyrus (Li et al., 2009). The negative correlation of rsFC-based decision scores with RAVLT measures shows that patients whose verbal learning and memory is more severely impaired exhibit a more distinctive rsFC pattern than patients with less impaired verbal memory.
format Online
Article
Text
id pubmed-5887536
institution National Center for Biotechnology Information
language English
publishDate 2018
publisher Oxford University Press
record_format MEDLINE/PubMed
spelling pubmed-58875362018-04-11 T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT Weiske, Johanna Ruef, Anne Haas, Shalaila Bonivento, Carolina Koutsouleris, Nikolaos Kambeitz-Ilankovic, Lana Schizophr Bull Abstracts BACKGROUND: Impairments in cognitive functioning are a core feature of psychotic disorders and they have been associated with resting-state functional connectivity (rsFC) alterations in patients suffering from psychosis (Dauverman et al., 2014). Multivariate pattern analysis (MVPA) has proven to be a useful tool in the investigation of rsFC alteration in psychosis and in detecting subtle differences in multidimensional data sets (Kambeitz et al., 2015). In this study, we differentiated recent-onset psychosis patients (ROP) from healthy controls (HC) using a Support Vector Machine (SVM) classification based on rsFC. Furthermore, we investigated the relationship of the discriminative rsFC pattern to neurocognitive measures. METHODS: Resting-state fMRI and neurocognitive measures were obtained from 220 HC and 115 ROP across 7 sites of the PRONIA consortium. The rsFC matrix was estimated for each subject by calculating pairwise correlations between mean time series of 90 brain regions based on AAL parcellation. A L1-regularized L2-loss SVM was trained to classify ROP vs. HC based on rsFC in a repeated nested cross-validation. Decision scores for ROP were correlated with cognitive measures derived from the following neuropsychological tests: Rey Auditory Verbal Learning Task (RAVLT), Phonetic and Semantic Verbal Fluency, Diagnostic Analysis of Nonverbal Accuracy, Forward and Backward Digit Span, Self-ordered Pointing Task, and Salience Attribution Test. RESULTS: The classification algorithm was able to differentiate ROP and HC with a balanced accuracy (BAC) of 71.3% based on rsFC. The discriminative connectivity pattern included short-range connections between left putamen and left hippocampus, right putamen and right caudate nucleus, left superior frontal and right inferior orbitofrontal regions, as well as long-range connections between left and right occipital cortex and left cingulate gyrus, left supramarginal gyrus and right temporal pole. Two negative correlations between the SVM decision scores for ROP and measures of the RAVLT were significant (delayed recall: r=-0.3, Bonferroni –adjusted p<.04; recall after interference: r=-0.3, Bonferroni-adjusted p<.02). DISCUSSION: The classification performance was driven by a rsFC pattern including areas involved in memory processing, such as hippocampus and cingulate gyrus (Allen et al., 2007) as well as regions related to language processing, such as the supra marginal gyrus (Li et al., 2009). The negative correlation of rsFC-based decision scores with RAVLT measures shows that patients whose verbal learning and memory is more severely impaired exhibit a more distinctive rsFC pattern than patients with less impaired verbal memory. Oxford University Press 2018-04 2018-04-01 /pmc/articles/PMC5887536/ http://dx.doi.org/10.1093/schbul/sby016.413 Text en © Maryland Psychiatric Research Center 2018. http://creativecommons.org/licenses/by/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
spellingShingle Abstracts
Weiske, Johanna
Ruef, Anne
Haas, Shalaila
Bonivento, Carolina
Koutsouleris, Nikolaos
Kambeitz-Ilankovic, Lana
T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT
title T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT
title_full T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT
title_fullStr T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT
title_full_unstemmed T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT
title_short T137. CLASSIFICATION OF RECENT-ONSET PSYCHOSIS BASED ON RESTING-STATE FUNCTIONAL CONNECTIVITY AND THE RELATIONSHIP TO NEUROCOGNITIVE IMPAIRMENT
title_sort t137. classification of recent-onset psychosis based on resting-state functional connectivity and the relationship to neurocognitive impairment
topic Abstracts
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5887536/
http://dx.doi.org/10.1093/schbul/sby016.413
work_keys_str_mv AT weiskejohanna t137classificationofrecentonsetpsychosisbasedonrestingstatefunctionalconnectivityandtherelationshiptoneurocognitiveimpairment
AT ruefanne t137classificationofrecentonsetpsychosisbasedonrestingstatefunctionalconnectivityandtherelationshiptoneurocognitiveimpairment
AT haasshalaila t137classificationofrecentonsetpsychosisbasedonrestingstatefunctionalconnectivityandtherelationshiptoneurocognitiveimpairment
AT boniventocarolina t137classificationofrecentonsetpsychosisbasedonrestingstatefunctionalconnectivityandtherelationshiptoneurocognitiveimpairment
AT koutsoulerisnikolaos t137classificationofrecentonsetpsychosisbasedonrestingstatefunctionalconnectivityandtherelationshiptoneurocognitiveimpairment
AT kambeitzilankoviclana t137classificationofrecentonsetpsychosisbasedonrestingstatefunctionalconnectivityandtherelationshiptoneurocognitiveimpairment