Imaging genetics is an emerging field dedicated to the study of genetic underpinnings of brain structure and function. Over the last decade, brain imaging techniques such as magnetic resonance imaging (MRI) have been increasingly applied to measure morphometry, task-based function and connectivity in living brains. Meanwhile, high-throughput genotyping employing genome-wide techniques has made it feasible to sample the entire genome of a substantial number of individuals. While there is growing interest in image-wide and genome-wide approaches which allow unbiased searches over a large range of variants, one of the most challenging problems is the correction for the huge number of statistical tests used in univariate models. In contrast, a reference-guided multivariate approach shows specific advantage for simultaneously assessing many variables for aggregate effects while leveraging prior information. It can improve the robustness of the results compared to a fully blind approach. In this dissertation we present a semi-blind multivariate approach, parallel independent component analysis with reference (pICA-R), to better reveal relationships between hidden factors of particular attributes. First, a consistency-based order estimation approach is introduced to advance the application of ICA to genotype data. The pICA-R approach is then presented, where independent components are extracted from two modalities in parallel and inter-modality associations are subsequently optimized for pairs of components. In particular, prior information is incorporated to elicit components of particular interests, which helps identify factors carrying small amounts of variance in large complex datasets. The pICA-R approach is further extended to accommodate multiple references whose interrelationships are unknown, allowing the investigation of functional influence on neurobiological traits of potentially related genetic variants implicated in biology. Applied to a schizophrenia study, pICA-R reveals that a complex genetic factor involving multiple pathways underlies schizophrenia-related gray matter deficits in prefrontal and temporal regions. The extended multi-reference approach, when employed to study alcohol dependence, delineates a complex genetic architecture, where the CREB-BDNF pathway plays a key role in the genetic factor underlying a proportion of variation in cue-elicited brain activations, which plays a role in phenotypic symptoms of alcohol dependence. In summary, our work makes several important contributions to advance the application of ICA to imaging genetics studies, which holds the promise to improve our understating of genetics underlying brain structure and function in healthy and disease.
","abstract_html":"<p>Imaging genetics is an emerging field dedicated to the study of genetic underpinnings of brain structure and function. Over the last decade, brain imaging techniques such as magnetic resonance imaging (MRI) have been increasingly applied to measure morphometry, task-based function and connectivity in living brains. Meanwhile, high-throughput genotyping employing genome-wide techniques has made it feasible to sample the entire genome of a substantial number of individuals. While there is growing interest in image-wide and genome-wide approaches which allow unbiased searches over a large range of variants, one of the most challenging problems is the correction for the huge number of statistical tests used in univariate models. In contrast, a reference-guided multivariate approach shows specific advantage for simultaneously assessing many variables for aggregate effects while leveraging prior information. It can improve the robustness of the results compared to a fully blind approach. In this dissertation we present a semi-blind multivariate approach, parallel independent component analysis with reference (pICA-R), to better reveal relationships between hidden factors of particular attributes. First, a consistency-based order estimation approach is introduced to advance the application of ICA to genotype data. The pICA-R approach is then presented, where independent components are extracted from two modalities in parallel and inter-modality associations are subsequently optimized for pairs of components. In particular, prior information is incorporated to elicit components of particular interests, which helps identify factors carrying small amounts of variance in large complex datasets. The pICA-R approach is further extended to accommodate multiple references whose interrelationships are unknown, allowing the investigation of functional influence on neurobiological traits of potentially related genetic variants implicated in biology. Applied to a schizophrenia study, pICA-R reveals that a complex genetic factor involving multiple pathways underlies schizophrenia-related gray matter deficits in prefrontal and temporal regions. The extended multi-reference approach, when employed to study alcohol dependence, delineates a complex genetic architecture, where the CREB-BDNF pathway plays a key role in the genetic factor underlying a proportion of variation in cue-elicited brain activations, which plays a role in phenotypic symptoms of alcohol dependence. In summary, our work makes several important contributions to advance the application of ICA to imaging genetics studies, which holds the promise to improve our understating of genetics underlying brain structure and function in healthy and disease.</p>","abstract_has_math":false,"creators":["Chen, Jiayu"],"institution":null,"degree_name":"Electrical Engineering","degree_level":"Dissertation","degree_discipline":"Electrical and Computer Engineering","degree_department":null,"school":null,"contributors":["Calhoun, Vince D.","Pattichis, Marios S.","Liu, Jingy","Perrone-Bizzozero, Nora I."],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2014,"date_issued":"2014-02-13T08:00:00Z","date_published":"2014-02-13T08:00:00Z","updated_at":"2026-07-24T05:27:10Z","subjects":["semi-blind","multivariate","ICA","neuroimaging","genomics","Electrical and Computer Engineering"],"languages":["English"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://digitalrepository.unm.edu/ece_etds/49","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Calhoun, Vince D.","Pattichis, Marios S.","Liu, Jingy","Perrone-Bizzozero, Nora I."]},{"key":"dc:creator","label":"Author","values":["Chen, Jiayu"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical and Computer Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Dissertation","Doctoral"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Electrical Engineering"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["semi-blind","multivariate","ICA","neuroimaging","genomics","Electrical and Computer Engineering"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["English"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalrepository.unm.edu/ece_etds/49"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Imaging genetics is an emerging field dedicated to the study of genetic underpinnings of brain structure and function. Over the last decade, brain imaging techniques such as magnetic resonance imaging (MRI) have been increasingly applied to measure morphometry, task-based function and connectivity in living brains. Meanwhile, high-throughput genotyping employing genome-wide techniques has made it feasible to sample the entire genome of a substantial number of individuals. While there is growing interest in image-wide and genome-wide approaches which allow unbiased searches over a large range of variants, one of the most challenging problems is the correction for the huge number of statistical tests used in univariate models. In contrast, a reference-guided multivariate approach shows specific advantage for simultaneously assessing many variables for aggregate effects while leveraging prior information. It can improve the robustness of the results compared to a fully blind approach. In this dissertation we present a semi-blind multivariate approach, parallel independent component analysis with reference (pICA-R), to better reveal relationships between hidden factors of particular attributes. First, a consistency-based order estimation approach is introduced to advance the application of ICA to genotype data. The pICA-R approach is then presented, where independent components are extracted from two modalities in parallel and inter-modality associations are subsequently optimized for pairs of components. In particular, prior information is incorporated to elicit components of particular interests, which helps identify factors carrying small amounts of variance in large complex datasets. The pICA-R approach is further extended to accommodate multiple references whose interrelationships are unknown, allowing the investigation of functional influence on neurobiological traits of potentially related genetic variants implicated in biology. Applied to a schizophrenia study, pICA-R reveals that a complex genetic factor involving multiple pathways underlies schizophrenia-related gray matter deficits in prefrontal and temporal regions. The extended multi-reference approach, when employed to study alcohol dependence, delineates a complex genetic architecture, where the CREB-BDNF pathway plays a key role in the genetic factor underlying a proportion of variation in cue-elicited brain activations, which plays a role in phenotypic symptoms of alcohol dependence. In summary, our work makes several important contributions to advance the application of ICA to imaging genetics studies, which holds the promise to improve our understating of genetics underlying brain structure and function in healthy and disease.
"]},{"key":"dc:title","label":"Title","values":["PARALLEL INDEPENDENT COMPONENT ANALYSIS WITH REFERENCE FOR IMAGING GENETICS: A SEMI-BLIND MULTIVARIATE APPROACH"]}]}],"canonical_facts":{"dc:contributor":["Calhoun, Vince D.","Pattichis, Marios S.","Liu, Jingy","Perrone-Bizzozero, Nora I."],"dc:creator":["Chen, Jiayu"],"dc:description.abstract":["Imaging genetics is an emerging field dedicated to the study of genetic underpinnings of brain structure and function. Over the last decade, brain imaging techniques such as magnetic resonance imaging (MRI) have been increasingly applied to measure morphometry, task-based function and connectivity in living brains. Meanwhile, high-throughput genotyping employing genome-wide techniques has made it feasible to sample the entire genome of a substantial number of individuals. While there is growing interest in image-wide and genome-wide approaches which allow unbiased searches over a large range of variants, one of the most challenging problems is the correction for the huge number of statistical tests used in univariate models. In contrast, a reference-guided multivariate approach shows specific advantage for simultaneously assessing many variables for aggregate effects while leveraging prior information. It can improve the robustness of the results compared to a fully blind approach. In this dissertation we present a semi-blind multivariate approach, parallel independent component analysis with reference (pICA-R), to better reveal relationships between hidden factors of particular attributes. First, a consistency-based order estimation approach is introduced to advance the application of ICA to genotype data. The pICA-R approach is then presented, where independent components are extracted from two modalities in parallel and inter-modality associations are subsequently optimized for pairs of components. In particular, prior information is incorporated to elicit components of particular interests, which helps identify factors carrying small amounts of variance in large complex datasets. The pICA-R approach is further extended to accommodate multiple references whose interrelationships are unknown, allowing the investigation of functional influence on neurobiological traits of potentially related genetic variants implicated in biology. Applied to a schizophrenia study, pICA-R reveals that a complex genetic factor involving multiple pathways underlies schizophrenia-related gray matter deficits in prefrontal and temporal regions. The extended multi-reference approach, when employed to study alcohol dependence, delineates a complex genetic architecture, where the CREB-BDNF pathway plays a key role in the genetic factor underlying a proportion of variation in cue-elicited brain activations, which plays a role in phenotypic symptoms of alcohol dependence. In summary, our work makes several important contributions to advance the application of ICA to imaging genetics studies, which holds the promise to improve our understating of genetics underlying brain structure and function in healthy and disease.
"],"dc:identifier":["https://digitalrepository.unm.edu/ece_etds/49"],"dc:language":["English"],"dc:subject":["semi-blind","multivariate","ICA","neuroimaging","genomics","Electrical and Computer Engineering"],"dc:title":["PARALLEL INDEPENDENT COMPONENT ANALYSIS WITH REFERENCE FOR IMAGING GENETICS: A SEMI-BLIND MULTIVARIATE APPROACH"],"thesis:degree_discipline":["Electrical and Computer Engineering"],"thesis:degree_level":["Dissertation","Doctoral"],"thesis:degree_name":["Electrical Engineering"]},"updated_at":"2026-07-24T05:27:10Z"}