Weighted Graph Comparison Techniques for Brain Connectivity Analysis by Basak Alper, Benjamin Bach, Nathalie Henry Riche.
Abstract:
The analysis of brain connectivity is a vast field in neuroscience with a frequent use of visual representations and an increasing need for visual analysis tools. Based on an in-depth literature review and interviews with neuroscientists, we explore high-level brain connectivity analysis tasks that need to be supported by dedicated visual analysis tools. A significant example of such a task is the comparison of different connectivity data in the form of weighted graphs. Several approaches have been suggested for graph comparison within information visualization, but the comparison of weighted graphs has not been addressed. We explored the design space of applicable visual representations and present augmented adjacency matrix and node-link visualizations. To assess which representation best support weighted graph comparison tasks, we performed a controlled experiment. Our findings suggest that matrices support these tasks well, outperforming node-link diagrams. These results have significant implications for the design of brain connectivity analysis tools that require weighted graph comparisons. They can also inform the design of visual analysis tools in other domains, e.g. comparison of weighted social networks or biological pathways.
The study used only eleven (11) participants on tasks that are domain dependent, but the authors are to be lauded for noticing:
While weighted graphs are present in a plethora of domains: computer networks, social networks, biological pathways networks, air traffic networks, commercial trade net-works; very few tools currently exist to represent and compare them. As we used generic comparison tasks during the study, our results can also inform the design of general weighted graph comparison tools.
Rather than inventing yet another weighted graph comparison tool, the authors compared some of the options for visualizing a weighted graph with users.
Evidence based interface design?
I first saw this at: Visualization Papers at CHI 2013 by Enrico Bertini.