Exploring relationships among topics in neuroscience literature using augmented reality visualization

Neuroscience researchers frequently consult scholarly articles to guide their experimental inquiries, primarily aiming to identify potential new experiments. To determine viable experiments, neuroscientists must understand established relationships among mental functions, brain regions, and neurological diseases. Since neuroscience experiments to verify specific relationships are costly and time-consuming, it is crucial to accurately identify those with the most potential to advance knowledge. For instance, a relationship mentioned in only a few studies might indicate a lack of experimental evidence rather than a non-existent causal link. We propose that using Augmented Reality (AR) to visualize data from neuroscience literature can help researchers identify existing or potentially fruitful relationships for further exploration. To support this assertion, we conducted three user-centered design studies using 3D AR visualizations tailored for different exploratory tasks. Our approach involved developing an immersive AR system named "DatAR," which formed the basis of our experimental studies. In the first study, we focused on identifying potential tasks for exploring neuroscience literature and determining suitable visual supports. We introduced an early version of DatAR to eight neuroscience students, who were tasked with identifying specific relationships. After completing the task, participants could explore the implemented functionalities and visualizations. Subsequent interviews validated the relationship-finding feature and the effectiveness of the visual representations. We later enhanced the system to display sentences and references linked to a specific disease and brain region relationship, supporting searches from a disease to related regions and vice versa. Experts in literature research evaluated this functionality, confirming that the DatAR prototype's relationship-finding feature was meaningful, understandable, and that the 3D visualizations aided in comprehending neuroscience topics. This provided initial evidence that AR visualizations could facilitate the identification of relationships conducive to fruitful experiments. Another observation from the initial studies was the frequent need among neuroscientists to compare how two diseases affect the same brain regions. In response, our third study explored how AR visualizations could assist in comparing affected brain regions. We provided 3D models emphasizing the regions impacted by different diseases, enhancing participants' understanding and enabling them to explore patterns and relationships more effectively. This approach allowed researchers to identify a small number of relevant papers for in-depth study without the need to review extensive literature. In subsequent evaluations, we assessed each developed functionality as a separate widget. The final study explored how these functionalities, when combined, could support neuroscience research. We engaged three neuroscientists to define representative tasks and scenarios that utilized most of the developed widgets in a coordinated manner. The results demonstrated that this ensemble of widgets was particularly beneficial for new researchers in neuroscience, helping them understand complex relationships between brain-related topics. The widgets supported each other in visualizing and verifying results, proving to be invaluable. Overall, our studies laid a solid foundation for understanding how AR can effectively elucidate complex relationships among neuroscience topics, demonstrating that an immersive AR environment can effectively display topics and their interconnections, thereby facilitating the exploration of neuroscience literature.