Integrating neurobiological markers of depression [Elektronische Ressource] : an fMRI-based pattern classification approach = Integration neurobiologischer Marker depressiver Erkrankungen mittels fMRT-basierter Musterklassifikation / submitted by Tim Hahn
121 pages
English

Découvre YouScribe en t'inscrivant gratuitement

Je m'inscris

Integrating neurobiological markers of depression [Elektronische Ressource] : an fMRI-based pattern classification approach = Integration neurobiologischer Marker depressiver Erkrankungen mittels fMRT-basierter Musterklassifikation / submitted by Tim Hahn

Découvre YouScribe en t'inscrivant gratuitement

Je m'inscris
Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus
121 pages
English
Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus

Description

Integrating neurobiological markers of depressionan: fMRI-based pattern classification approach Integration neurobiologischer Marker depressiver Ekrankungen mittels fMRT-basierter Musterklassifikation Doctoral thesis for a doctoral degree at the Graduate School of Life Sciences, Julius-Maximilians-Universität Würzburg, SectionN eurosciencesubmitted by Tim Hahn from Lennestadt Würzburg 2010 Submitted on: April 1, 2010 Members of the Promotionskomitee: Chairperson: Prof. Dr. Dr. Martin J. Müller Primary Supervisor: Prof. Dr. Andreas J. Fallgatter Supervisor (Second): Prof. Dr. Klaus-Peter Lesch Supervisor (Third): Prof. Dr. Martin A. Heisenberg Date of Public Defence: July 26, 2010 Date of Receipt of Certificates: “Prediction is very difficult, especially abouuret .t”h e futNiels Bohr (1885-1962) Table of Contents 0. Abstract..............................................................7.........1. Introduction......................................................1.1..........2. Part I – Integrating biomarkers: development of a lmtiu-source pattern classification algorithm .............................................1.4...............2.1. Approaches to classification....................................................................... 162.2. Goals and challenges of algorithm developmen.t.................................... 202.3. Algorithm development...................................2.2........................2.3.1. First-level prediction.

Sujets

Informations

Publié par
Publié le 01 janvier 2010
Nombre de lectures 7
Langue English

Extrait

Integrating neurobiological markers of depressionan:
fMRI-based pattern classification approach
Integration neurobiologischer Marker depressiver Ekrankungen
mittels fMRT-basierter Musterklassifikation
Doctoral thesis for a doctoral degree
at the Graduate School of Life Sciences,
Julius-Maximilians-Universität Würzburg,
SectionN euroscience
submitted by
Tim Hahn
from Lennestadt
Würzburg
2010 Submitted on: April 1, 2010
Members of the Promotionskomitee:
Chairperson: Prof. Dr. Dr. Martin J. Müller
Primary Supervisor: Prof. Dr. Andreas J. Fallgatter
Supervisor (Second): Prof. Dr. Klaus-Peter Lesch
Supervisor (Third): Prof. Dr. Martin A. Heisenberg
Date of Public Defence: July 26, 2010
Date of Receipt of Certificates: “Prediction is very difficult, especially abouuret .t”h e fut
Niels Bohr (1885-1962) Table of Contents
0. Abstract..............................................................7.........
1. Introduction......................................................1.1..........
2. Part I – Integrating biomarkers: development of a lmtiu-source pattern
classification algorithm .............................................1.4...............
2.1. Approaches to classification....................................................................... 16
2.2. Goals and challenges of algorithm developmen.t.................................... 20
2.3. Algorithm development...................................2.2........................
2.3.1. First-level prediction...............................2.3..........................
2.3.2. Second-level prediction....................................................... 31
2.3.3. Significance Testing...................................................... 35
2.3.4. Multivariate feature mapping............................................................... 36
2.4. Summary..................................................................................4.0.................
3. Part II – Classification in the context of d.ep.r.e.ssi..o.n................................... 42
3.1. Introduction ..............................................4.2................
3.1.1. The concept of depression................................................... 43
3.1.1.1. Epidemiology of depression......................................... 43
3.1.1.2. Symptoms and diagnosis of depression....................................... 45
3.1.2. Biological markers of depression........................................................ 46
3.1.2.1. Processing of emotional stimuli..................................... 47
3.1.2.2. Neuroimaging markers................................................ 50
3.1.2.3. Other biological markers.............................................. 56
3.2. Summary and goals of the study.................................................... 58
3.3. Materials and Methods.....................................................6.3.........................
3.3.1. Participants..................................................................6.3......................
3.3.2. Tasks and procedures........................................................... 65
3.3.3. Functional Magnetic Resonance Imaging................................. 68
3.3.4. Algorithm application.............................7.3..........................
3.4. Results......................................................7.4...............
3.4.1. Classification based on single biomarker.s................................. 74
3.4.2. Integrated biomarker classification...................................................... 75
3.4.3. Multivariate spatial mapping of neural pro.ce.sse.s............................ 77
4. Discussion........................................................7.9...............
4.1. Single biomarkers of depression................................................... 80
4.2. Combining symptom-related biomarkers of depressio.n.......................... 83
4.3. Methodological considerations..................................................... 89
4.4. Limitations..................................................................................9.2...............
4.5. Future directions...........................................9.8...................
5. References.......................................................1.0.2..............
6. Appendix......................................................1.1.1............7
0. Abstract
English
While depressive disorders are, to date, diagnosed bda soen behavioral
symptoms and course of illness, the interest in neurlobgiocal markers of psychiatric
disorders has grown substantially in recent years. How,e vercurrent classification
approaches are mainly based on data from a single mbiarkeor, making it difficult to
predict diseases such as depression which are charactebriy zae dc omplex pattern of
symptoms. Accordingly, none of the previously itendv esistingglea biomarkers has
shown sufficient predictive power for practicaonl .a p plicati
In this work, we therefore propose an algorithm wnhtegiratches n euiroimaging
data associated with multiple, symptom-related nperuorcale sses relevant in
depression to improve classification accuracy. First, wtei ifideend the core-symptoms
of depression from standard classification systems. Thee nd,esi wgned and
conducted three experimental paradigms probing psychogiolcal processes known to
be related to these symptoms using functional MagnetiResoc nance Imaging. In
order to integrate the resulting 12 high-dimbienosimoanrkearl s, we developed a
multi-source pattern recognition algorithm based onm ba icnation of Gaussian
Process Classifiers and decision trees.
Applying this approach to a group of 30 healths y aconndt r3ol0 depressive in-
patients who were on a variety of medications and dyieds pvlarying degrees of
symptom-severity allowed for high-accuracy single-subjclectassi fication. Specifically,
integrating biomarkers yielded an accuracy of 83% wh thiel ebest of the 12 single
biomarkers alone classified a significantly lower noufm bsuerbjects (72%) correctly.
Thus, integrated biomarker-based classification ofte roa gheneous, real-life
sample resulted in accuracy comparable to the highest erv eachieved in previous 8
single biomarker research. Furthermore, investigatfi tohn e ofinal prediction model
revealed that neural activation during the proc essiofn gneutral facial expressions,
large rewards, and safety cues is most relevant for over- all classification. We
conclude that combining brain activation trhelaet edc otroe -symptoms of depression
using the multi-source pattern classification approvealochped die n this work
substantially increases classification accuracy while pronvgi ad isparse relational
biomarker-model for future prediction. 9
Deutsch
Während depressive Erkrankungen bislang größtenteils auf der Basis von
Symptomen auf der Verhaltensebene und den jeweiligen Krankheitsverläufen
diagnostiziert werden, hat das Interesse an dere nVderuwng neurobiologischer
Marker bei psychischen Erkrankungen in den letzten Jahstraernk zugenommen. Da
jedoch die momentan verfügbaren Klassifikationsansätuzem ezist auf Informationen
eines einzelnen Biomarkers beruhen, ist die Vorhervosagne auf der Symptomebene
so komplexen Erkrankungen wie Depressionen in der Praxids eutlich erschwert.
Dementsprechend konnte keiner der einzelnen bisher uernsut chten Biomarker eine
Vorhersagegüte erreichen, die für die praktische Aungw eenidnes solchen Ansatzes
im klinischen Alltag ausreichend wäre.
Vor diesem Hintergrund schlagen wir deshalb zur Verbesserung der
Klassifikationsgüte einen Algorithmus vor, der Messda tvenielfältiger
depressionsrelevanter neuronaler Prozesse integriert. Zunächst wurden hierzu die
Kernsymptome depressiver Erkrankungen aus standardisierten
Klassifikationssystemen ermittelt. Anschließend entwi ckewlitenr drei experimentelle
Paradigmen, welche die Messung neuronaler Korrelate der mit den depressiven
Kernsymptomen assoziierten psychologischen Prozesse mls ifutntektioneller
Kernspintomographie ermöglichen. Um die resunlti 1e2re nhdeochdimensionalen
Biomarker zu integrieren, entwickelten wir basieurefn dd ear Kombination von Gauß-
Prozess Klassifikatoren und Entscheidungsbäumen einen eziwstufigen
Mustererkennungsalgorithmus für multiple, hochdimenlsie oDanteanquellen.
Dieser Ansatz wurde an einer Gruppe von 30 gesunden Pro banden und 30
unterschiedlich schwer betroffenen und unterschiedlichzi ermteend stationären
depressiven Patienten evaluiert. Insgesamt ermöglicht dre Ansatz eine hohe
Klassifikationsgüte auf Einzelfallebene. Insbesonderee dIintegration der 10
verschiedenen Biomarker führte zu einer Klassifikatioe nvsognü t83%, wohingegen
die alleinige Klassifikationsgüte der 12 einozmelnarekenr mBit bestenfalls 72%
deutlich geringer ausfiel.
Somit konnte der entwickelte Klassifikationsansatz niern heieterogenen, im Alltag
aber typisch anzutreffenden depressiven Patientenstich bpe,ro eine
Klassifikationsgüte erreichen, die mit der bislagnlgi bchestmenö durch einzelne
Biomarker erreichten Klassifikationsgüte in selektivene lEstichnz proben vergleichbar
ist. Darüber hinaus zeigte die Analyse des empirirschä

  • Univers Univers
  • Ebooks Ebooks
  • Livres audio Livres audio
  • Presse Presse
  • Podcasts Podcasts
  • BD BD
  • Documents Documents