Data-driven Analysis for Natural Studies in Functional Brain Imaging

Jarkko Ylipaavalniemi

Data-driven Analysis for Natural Studies in Functional Brain Imaging

Julkaisun otsikon käännös: Aineistolähtöinen analysointi luonnollisille koeasetelmille toiminnallisessa aivokuvantamisessa

Jarkko Ylipaavalniemi

Tutkimustuotos: Doctoral Thesis › Collection of Articles

Abstrakti

Toiminnallisesta magneettiresonassikuvantamisesta (fMRI) on tullut tehokas työkalu neu- rotieteessä ihmisaivojen toiminnan kartoittamiseen. Yleensä fMRI-menetelmän yhteydessä käytetään melko yksinkertaista ärsykkeiden sarjaa, jolla pyritään hyvään signaali-kohinasuh- teeseen tilastollista hypoteesitestausta varten. Kun siirrytään kohti luonnollisia ärsykkeitä, tällaiset yksinkertaiset koeasetelmat eivät ole enää päteviä. Tämän väitöstyön tavoite on ollut kehittää uusia aineistolähtöisiä menetelmiä luonnollisten ärsykkeiden aiheuttamien aivovas- teiden luotettavaan tunnistamiseen. Aina 1800-luvun alusta lähtien neurotieteessä on vallinnut käsitys siitä, että erilliset aivoalu- eet vastaavat tiettyjä neuropsykologisia toimintoja. Moderni tutkimus on kuitenkin osoittanut, että monet toiminnot perustuvat hajautettuihin aivoalueiden verkostoihin ja kukin aivoalue voi toimia osana useampaa verkostoa. Puhtaasti hypoteesitestaukseen perustuvat menetelmät ovat olleet hyvin suosittuja aivokuvantamisen yhteydessä. Kun tutkimus keskittyy yhä enem- män toiminnallisen erikoistumisen sijaan vuorovaikutuksiin perustuviin verkostoihin, tällaisia menetelmiä ei voida enää käyttää. Niiden sijaan modernit koneoppimismenetelmät mahdollis- tavat puhtaasti aineistolähtöisen tavan mittausten selittämiseen. Nämä menetelmät eivät tar- vitse tietoa ärsykkeistä eivätkä tee oletuksia siitä ovatko mittaustulokset ärsykkeistä riippu- vaisia. Mittausten oletetaan muodostuvan usean aktivaation sekoituksista, jotka voivat sisältää monia sijainniltaan hajautettuja prosesseja, sekä mittausvirheitä. Jokainen prosessi voidaan selittää komponenttina, jolla on muista eriävä aikasarja ja sijaintikartta. Väitöskirjan pääasiallinen tulos on luotettava riippumattomien komponenttien menetelmä (ICA), joka on vapaasti saatavilla Arabica-ohjelmistossa. Uuden menetelmän toimivuutta on testattu usealla fMRI-aineistolla ja osoitettu, että menetelmä kykenee selittämään mittaukset uudella tavalla ja tuottamaan tietoa aineistosta, joka ei muuten olisi mahdollista. Menetelmän toimivuus on todistettu myös teoreettisen tilastotieteen keinoin, joka tarjoaa myös kattavat perusteet menetelmän käytännön hyödyntämiseen. Menetelmää myös jatkokehitettiin hajau- tettujen verkostojen analysointiin sopivaksi yhdistämällä se kanonisen korrelaatioanalyysin (CCA) kanssa. Tämän laajennuksen osoitettiin soveltuvan luonnollisia ärsykkeitä käyttävien fMRI-aineistojen analysointiin. Lopuksi menetelmää parannettiin myös sellaisia tilanteita var- ten, joissa havaitaan riippumattomia aliavaruuksia. Näin tapahtuu usein käytännössä, sillä aineiston ei voida taata täyttävän kaikkia menetelmissä tehtyjä oletuksia.

Julkaisun otsikon käännös	Aineistolähtöinen analysointi luonnollisille koeasetelmille toiminnallisessa aivokuvantamisessa
Alkuperäiskieli	Englanti
Pätevyys	Tohtorintutkinto
Myöntävä instituutio	Aalto-yliopisto
Valvoja/neuvonantaja	Oja, Erkki, Vastuuprofessori Vigário, Ricardo, Ohjaaja, Ulkoinen henkilö
Kustantaja	Aalto University
Painoksen ISBN	978-952-60-5136-9
Sähköinen ISBN	978-952-60-5137-6
Tila	Julkaistu - 2013
OKM-julkaisutyyppi	G5 Artikkeliväitöskirja

Tutkimusalat

aliavaruudet
toiminnallinen magneettiresonanssikuvantaminen
funktionaaliset verkostot
funktionaaliset yksiköt
kanoninen korrelaatioanalyysi
luonnolliset ärsykkeet
luotettavuus
riippumattomien komponenttien menetelmä

Pääsy asiakirjaan

http://urn.fi/URN:ISBN:978-952-60-5137-6

Siteeraa tätä

@phdthesis{2608b6a8871049bb87fd154d63ecc7b0,

title = "Data-driven Analysis for Natural Studies in Functional Brain Imaging",

abstract = "In neuroscience, functional magnetic resonance imaging (fMRI) has become a powerful tool in human brain mapping. Typically, fMRI is used with a rather simple stimulus sequence, aiming at improving signal-to-noise ratio for statistical hypothesis testing. When natural stimuli are used, the simple designs are no longer appropriate. The aim of this thesis is in developing data-driven approaches for reliable inference of brain correlates to natural stimuli. Since the beginning of the nineteenth century, neuroscience has focused on the idea that distinct regions of the brain support particular mental processes. However, modern research recognizes that many functions rely on distributed networks, and that a single brain region may participate in more than one function. These rapid paradigm changes in neuroscience raise important methodological challenges. Purely hypothesis-driven methods have been used extensively in functional imaging studies. As the focus in brain research is shifting away from functional specialization towards interaction-based functional networks, those approaches are no longer appropriate. In contrast to the classic statistical hypothesis testing approaches, modern machine learning methods allow for a purely data-driven way to describe the data. They do not use the stimuli, and make no assumptions about whether the brain processes are stimulus related or not. The recordings for each brain region may contain a complicated mixture of activity, which is produced by many spatially distributed processes, and artifacts. Each process can be described as a component having a separate time series and spatial extent, and producing simultaneous changes in the fMRI signals of many regions. The main contribution of the thesis is a reliable independent component analysis (ICA) approach, which is available in the Arabica toolbox. The usefulness of the approach was tested extensively with fMRI data, showing that the method is capable of providing insights into the data that would not be attainable otherwise. The new method was also theoretically analyzed and its asymptotic convergence was proven. The theory offers a thorough explanation of how the method works and justifies its use in practice. Then, the new method is further developed for analyzing networks of distributed brain activity, by combining it with canonical correlation analysis (CCA). The extension was shown to be particularly useful with fMRI studies that use natural stimuli. The approach is further extended to be applicable in cases where independent subspaces emerge, which often happens when using real measurement data that is not guaranteed to fit all the assumptions made in the development of the methods.",

keywords = "canonical correlation analysis, independent component analysis, functional elements, functional networks, functional magnetic resonance imaging, natural stimulation, subspaces, reliability, aliavaruudet, toiminnallinen magneettiresonanssikuvantaminen, funktionaaliset verkostot, funktionaaliset yksik{\"o}t, kanoninen korrelaatioanalyysi, luonnolliset {\"a}rsykkeet, luotettavuus, riippumattomien komponenttien menetelm{\"a}, canonical correlation analysis, independent component analysis, functional elements, functional networks, functional magnetic resonance imaging, natural stimulation, subspaces, reliability",

author = "Jarkko Ylipaavalniemi",

year = "2013",

language = "English",

isbn = "978-952-60-5136-9",

series = "Aalto University publication series DOCTORAL DISSERTATIONS",

publisher = "Aalto University",

number = "70",

address = "Finland",

school = "Aalto University",

}

TY - BOOK

T1 - Data-driven Analysis for Natural Studies in Functional Brain Imaging

AU - Ylipaavalniemi, Jarkko

PY - 2013

Y1 - 2013

N2 - In neuroscience, functional magnetic resonance imaging (fMRI) has become a powerful tool in human brain mapping. Typically, fMRI is used with a rather simple stimulus sequence, aiming at improving signal-to-noise ratio for statistical hypothesis testing. When natural stimuli are used, the simple designs are no longer appropriate. The aim of this thesis is in developing data-driven approaches for reliable inference of brain correlates to natural stimuli. Since the beginning of the nineteenth century, neuroscience has focused on the idea that distinct regions of the brain support particular mental processes. However, modern research recognizes that many functions rely on distributed networks, and that a single brain region may participate in more than one function. These rapid paradigm changes in neuroscience raise important methodological challenges. Purely hypothesis-driven methods have been used extensively in functional imaging studies. As the focus in brain research is shifting away from functional specialization towards interaction-based functional networks, those approaches are no longer appropriate. In contrast to the classic statistical hypothesis testing approaches, modern machine learning methods allow for a purely data-driven way to describe the data. They do not use the stimuli, and make no assumptions about whether the brain processes are stimulus related or not. The recordings for each brain region may contain a complicated mixture of activity, which is produced by many spatially distributed processes, and artifacts. Each process can be described as a component having a separate time series and spatial extent, and producing simultaneous changes in the fMRI signals of many regions. The main contribution of the thesis is a reliable independent component analysis (ICA) approach, which is available in the Arabica toolbox. The usefulness of the approach was tested extensively with fMRI data, showing that the method is capable of providing insights into the data that would not be attainable otherwise. The new method was also theoretically analyzed and its asymptotic convergence was proven. The theory offers a thorough explanation of how the method works and justifies its use in practice. Then, the new method is further developed for analyzing networks of distributed brain activity, by combining it with canonical correlation analysis (CCA). The extension was shown to be particularly useful with fMRI studies that use natural stimuli. The approach is further extended to be applicable in cases where independent subspaces emerge, which often happens when using real measurement data that is not guaranteed to fit all the assumptions made in the development of the methods.

AB - In neuroscience, functional magnetic resonance imaging (fMRI) has become a powerful tool in human brain mapping. Typically, fMRI is used with a rather simple stimulus sequence, aiming at improving signal-to-noise ratio for statistical hypothesis testing. When natural stimuli are used, the simple designs are no longer appropriate. The aim of this thesis is in developing data-driven approaches for reliable inference of brain correlates to natural stimuli. Since the beginning of the nineteenth century, neuroscience has focused on the idea that distinct regions of the brain support particular mental processes. However, modern research recognizes that many functions rely on distributed networks, and that a single brain region may participate in more than one function. These rapid paradigm changes in neuroscience raise important methodological challenges. Purely hypothesis-driven methods have been used extensively in functional imaging studies. As the focus in brain research is shifting away from functional specialization towards interaction-based functional networks, those approaches are no longer appropriate. In contrast to the classic statistical hypothesis testing approaches, modern machine learning methods allow for a purely data-driven way to describe the data. They do not use the stimuli, and make no assumptions about whether the brain processes are stimulus related or not. The recordings for each brain region may contain a complicated mixture of activity, which is produced by many spatially distributed processes, and artifacts. Each process can be described as a component having a separate time series and spatial extent, and producing simultaneous changes in the fMRI signals of many regions. The main contribution of the thesis is a reliable independent component analysis (ICA) approach, which is available in the Arabica toolbox. The usefulness of the approach was tested extensively with fMRI data, showing that the method is capable of providing insights into the data that would not be attainable otherwise. The new method was also theoretically analyzed and its asymptotic convergence was proven. The theory offers a thorough explanation of how the method works and justifies its use in practice. Then, the new method is further developed for analyzing networks of distributed brain activity, by combining it with canonical correlation analysis (CCA). The extension was shown to be particularly useful with fMRI studies that use natural stimuli. The approach is further extended to be applicable in cases where independent subspaces emerge, which often happens when using real measurement data that is not guaranteed to fit all the assumptions made in the development of the methods.

KW - canonical correlation analysis

KW - independent component analysis

KW - functional elements

KW - functional networks

KW - functional magnetic resonance imaging

KW - natural stimulation

KW - subspaces

KW - reliability

KW - aliavaruudet

KW - toiminnallinen magneettiresonanssikuvantaminen

KW - funktionaaliset verkostot

KW - funktionaaliset yksiköt

KW - kanoninen korrelaatioanalyysi

KW - luonnolliset ärsykkeet

KW - luotettavuus

KW - riippumattomien komponenttien menetelmä

KW - canonical correlation analysis

KW - independent component analysis

KW - functional elements

KW - functional networks

KW - functional magnetic resonance imaging

KW - natural stimulation

KW - subspaces

KW - reliability

M3 - Doctoral Thesis

SN - 978-952-60-5136-9

T3 - Aalto University publication series DOCTORAL DISSERTATIONS

PB - Aalto University

ER -

Data-driven Analysis for Natural Studies in Functional Brain Imaging

Abstrakti

Tutkimusalat

Pääsy asiakirjaan

Sormenjälki

Siteeraa tätä