Poster No:
1703
Submission Type:
Abstract Submission
Authors:
Imre Kertesz1, Stefan Fraessle1, Jakob Heinzle1, Klaas Stephan1,2
Institutions:
1Translational Neuromodeling Unit (TNU), University of Zurich & ETH Zurich, Zurich, Switzerland, 2Max Planck Institute for Metabolism Research, Cologne, Germany
First Author:
Imre Kertesz
Translational Neuromodeling Unit (TNU), University of Zurich & ETH Zurich
Zurich, Switzerland
Co-Author(s):
Stefan Fraessle
Translational Neuromodeling Unit (TNU), University of Zurich & ETH Zurich
Zurich, Switzerland
Jakob Heinzle
Translational Neuromodeling Unit (TNU), University of Zurich & ETH Zurich
Zurich, Switzerland
Klaas Stephan
Translational Neuromodeling Unit (TNU), University of Zurich & ETH Zurich|Max Planck Institute for Metabolism Research
Zurich, Switzerland|Cologne, Germany
Introduction:
Mapping effective (directed) connectivity in whole-brain networks is a key challenge in Computational Neuroscience and Computational Psychiatry. Using neuroimaging data, dynamic causal modeling (DCM) is a popular framework to estimate brain connectivity. Due to its computational complexity, analysis is restricted to small networks with few regions. A novel approach called regression DCM (rDCM) overcomes those computational limitations, rendering inference on effective connectivity in whole-brain networks feasible.
Methods:
In order to turn the novel approach into a beneficial tool in the context of computational assays, it is indispensable to have computational runtimes that are compatible within clinical practice. Motivated by this desideratum, we present an open-source software package in Julia that implements the rDCM framework. The new Julia language endorses the development of highly efficient code, resulting in minimal resource usage. The robust and user-friendly implementation allows to perform whole-brain analysis with only few lines of code. With Julia being a freely available open-source project, we hope to reach a large audience.
Results:
Here we illustrate the utility of the package, present runtime and memory consumption comparison between the Julia and Matlab implementation. Furthermore, we outline the typical workflow when using the new tool and demonstrate it in application to concrete examples.
Conclusions:
Using the new Julia language, we implemented a package that allows to perform whole-brain effective connectivity analysis in the order of seconds to minutes on standard hardware.
This new tool is part of the Translational Algorithms for Psychiatry-Advancing Science (TAPAS) software package (https://translationalneuromodeling.github.io/tapas/) – an open-source collection of building blocks for computational assays in psychiatry developed at the Translational Neuromodeling Unit (TNU).
Modeling and Analysis Methods:
Bayesian Modeling
Connectivity (eg. functional, effective, structural) 2
fMRI Connectivity and Network Modeling 1
Keywords:
Computational Neuroscience
FUNCTIONAL MRI
Modeling
Open-Source Code
Open-Source Software
Other - Bayesian modeling
1|2Indicates the priority used for review
Provide references using author date format
Fraessle, S. (2017), 'Regression DCM for fMRI', NeuroImage, vol. 155, pp. 406-421
Fraessle, S. (2018), 'A generative model of whole-brain effective connectivity', NeuroImage, vol. 179, pp. 505-529
Fraessle, S. (2021), 'TAPAS: An Open-Source Software Package for Translational', Neuromodeling and Computational Psychiatry, vol. 12
Friston, K. (2003), 'Dynamic causal modelling’, NeuroImage, vol. 19, no. 4, pp. 1273-1302