UPDATE: New Paper 'Interpreting null models of resting-state functional MRI dynamics:not throwing the model out with the hypothesis'

Abstract: Null models are useful for assessing whether a dataset exhibits a non-trivial property of interest. These models have recently gained interest in the neuroimaging community as means to explore dynamic properties of functional Magnetic Resonance Imaging (fMRI) time series. Interpretation of null-model testing in this context may not be straightforward because (i) null hypotheses associated to different null models are sometimes unclear and (ii) fMRI metrics might be ‘trivial’, i.e. preserved under the null hypothesis, and still be useful in neuroimaging applications. In this commentary, we review several commonly used null models of fMRI time series and discuss the interpretation of the corresponding tests. We argue that, while null-model testing allows for a better characterization of the statistical properties of fMRI time series and associated metrics, it should not be considered as a mandatory validation step to assess their relevance in representing brain functional dynamics.
https://www.sciencedirect.com/science/article/pii/S1053811921007916

UPDATE: New Paper on 'Integrating Prior Knowledge in Mixed Initiative Social Network Clustering'

We propose a new approach-called PK-clustering-to help social scientists create meaningful clusters in social networks. Many clustering algorithms exist but most social scientists find them difficult to understand, and tools do not provide any guidance to choose algorithms, or to evaluate results taking into account the prior knowledge of the scientists. Our work introduces a new clustering approach and a visual analytics user interface that address this issue. It is based on a process that 1) captures the prior knowledge of the scientists as a set of incomplete clusters, 2) runs multiple clustering algorithms (similarly to clustering ensemble methods), 3) visualizes the results of all the algorithms ranked and summarized by how well each algorithm matches the prior knowledge, 4) evaluates the consensus between user-selected algorithms and 5) allows users to review details and iteratively update the acquired knowledge. We describe our approach using an initial functional prototype, then provide two examples of use and early feedback from social scientists. We believe our clustering approach offers a novel constructive method to iteratively build knowledge while avoiding being overly influenced by the results of often randomly selected black-box clustering algorithms.
https://ieeexplore.ieee.org/document/9237999
https://hal.inria.fr/hal-02566438
https://aviz.fr/pkclustering

UPDATE: New Paper on 'Interpreting Null Models of Resting-State Functional MRI'

Null models are necessary for assessing whether a dataset exhibits non-trivial statistical properties. These models have recently gained interest in the neuroimaging community as means to explore dynamic properties of functional Magnetic Resonance Imaging (fMRI) time series. Interpretation of null-model testing in this context may not be straightforward because (i) null hypotheses associated to different null models are sometimes unclear and (ii) fMRI metrics might be `trivial', i.e. preserved under the null hypothesis, and still be useful in neuroimaging applications. In this commentary, we review several commonly used null models of fMRI time series and discuss the interpretation of the corresponding tests. We argue that, while null-model testing allows for a better characterization of the statistical properties of fMRI time series and associated metrics, it should not be considered as a mandatory validation step to assess their relevance in neuroimaging applications.
https://www.biorxiv.org/content/10.1101/2021.03.30.437514v1

2020-09-03: Creativity workshop on EEG and Epilepsy

Six clinicians from the Geneva University Hospital, including medical doctors and epilepsy experts, participated in the ‘Creativity workshop on EEG and Epilepsy’. Through several activities guided by workshop facilitators, we identified several key elements where novel modelling and visualization tools could lead to better treatment of epilepsy. These included, e.g., better description of differences and commonalities between different forms of epilepsy, and more intuitive ways to navigate the complex spatio-temporal space of electroencephalography (EEG) data.

UPDATE: Paper accepted to BELIV 2020

Our paper on 'Extending Recommendations for Creative Visualization-Opportunities Workshops’ has been accepted to BELIV 2020 'Evaluation and Beyond - Methodological Approaches for Visualization'.

Abstract: Participatory design is an approach in human-computer interaction that involves all relevant stakeholders coequally in the design process. A recent participatory method for visualization design is the creative visualization-opportunities (CVO) workshop, which is used to efficiently develop visualization design requirements in the early stages of applied visualization work. In this paper we report on our experiences of running four CVO workshops in different domains with diverse participants to explore new methods and variations of workshop variables. Through reflection on our experiences, we propose two contributions that extend existing guidance for planning, executing, and analyzing CVO workshops: a set of 12 pragmatic recommendations that extend and complement existing ones; and a recommended method for analyzing workshop results, called user stories. Additionally, we report on the outcomes of our successful workshops to provide evidence for the efficacy of CVO workshops.
https://ieeexplore.ieee.org/document/9307763

UPDATE: Paper accepted in IEEE Signal Processing Magazine

The paper 'Community-Aware Graph Signal Processing' by Miljan Petrovic, Raphael Liegeois, Thomas A. W. Bolton, Dimitri Van De Ville has just been accepted for publication in the IEEE SPM Special Issue on Graph Signal Processing: Foundations and Emerging Directions.

Abstract: Graph signal processing (GSP) extends usual signal processing tools to data living on graphs. This is classically done by exploiting the structure of the graph Laplacian which encodes diffusive properties. In this work, in contrast, we define GSP based on the graph modularity matrix which encodes the graph community structure thereby making GSP 'community aware'. Methodological challenges coming with this new definition are first discussed, and applications are then presented.
https://arxiv.org/abs/2008.10375

UPDATE: Paper accepted in Network Neuroscience

Our paper on 'Revisiting correlation-based functional connectivity and its relationship with structural connectivity' has just been accepted in Network Neuroscience.

Abstract: In network neuroscience, functional connectivity (FC) denotes statistical dependencies between brain function in different regions. It is classically evaluated from pairwise correlations between functional time series. This FC measure captures both ‘direct’ and ‘indirect’ statistical dependencies which makes the comparison to structural connectivity (SC), which encodes only ‘direct’ anatomical connections, of limited relevance. In this paper we revisit the use of alternative FC measures yielding more natural SC-FC comparisons. In particular, the precision matrix, defined as the inverse of the correlation matrix, is shown to provide a better SC-FC match as compared to correlation-based FC. Then, using a simple model of brain structure and function interactions, we show that the SC-FC match can also be used to explore different aspects of brain anatomy and function.

2019-10-04: Creativity Workshop in Paris

Eight researchers from the field of digital humanities participated in our creativity workshop in Paris. They are working on the analysis of historic contract data. In this full-day workshop, we wanted to gather important information on how we can support the analysts in their daily work through visualization. This process consisted of a set of consecutive activities guided by workshop facilitators.

UPDATE: Visualization of Dynamic Network Clusterings

We have designed a visual representation for medium-sized (about 200-1000 vertices and 10-200 time steps) dynamic networks that support aggregation in time and topology. We are currently improving it to scale to larger sizes, but we are targeting around 10,000 vertices or a bit more depending on the amount of clustering we can do. These visualizations will be soon available on our tool at https://aviz.fr/paohvis.

2019-4-3: Chist-ERA meeting in Budapest

The IVAN team presented the project's first year developments at the annual CHIST-ERA meeting in Bucharest on 3rd-4th April, 2019. The next annual meeting will be held in Prague in April 2020.

UPDATE: New paper on "Slepian"

We have developed a new spectral representation of graphs based on the Slepian framework. By combining minimization of embedded distance and maximization of information concentration, this decomposition offers a novel way to explore networks that allows to zoom in specific subnetworks such as specific social networks communities or brain networks. You can find the Matlab code for the analysis presented in the paper on https://c4science.ch/source/guidedGSE/.