Understanding the human brain's complex structure and intricate interconnections is crucial for neuroscience research. Traditional 2D representations, while useful, struggle to effectively convey the three-dimensional complexity of brain anatomy and the pathways connecting different structures. The increasing focus on brain connectivity necessitates more intuitive visualization tools. While several AR/VR solutions exist for neuroanatomical education, they primarily concentrate on anatomy alone, lacking the interactive exploration and detailed explanations needed to fully grasp complex neural pathways and their functions. This research aims to address this gap by creating SONIA (immerSive custOmizable Neuro learnIng plAform), a user-friendly VR system. SONIA incorporates a multi-scale interaction paradigm inspired by VR-based geological data navigation, allowing users to interact with both large-scale and small-scale brain models. A progression-based learning strategy uses completion metrics and multimedia elements (visual guidance and audio voice-overs) to create a stimulating and enriching user experience. The system's customizable design enables easy incorporation of various learning materials, including detailed narratives of brain subsystems, opening doors for numerous future applications. The study utilizes a specific example—the anxiety-related functional brain network—to test the functionality and effectiveness of SONIA.

Existing AR/VR solutions have shown promise in improving the understanding of neuroanatomy. However, few systems address the visualization and demonstration of neural pathways and brain networks comprehensively. NeuroCave, a web-based platform, allows exploration of connectomic data, while other workflows leverage existing software to visualize tractograms and functional connectivity. Recent applications focus on specific pathways, like the visual and auditory systems. While some address the challenges of 3D visualization, few experiment with novel interaction paradigms that enhance usability and learning. Moreover, none of the existing systems integrate descriptive insights along the pathway exploration or incorporate well-designed learning modules. SONIA aims to overcome these limitations by providing interactive visualizations and learning modules for both neuroanatomy and associated structural and functional networks.

SONIA's virtual brain model was constructed using the AAL116 brain atlas, with additional manual segmentation for the bed nucleus of the stria terminalis (BNST). The VR environment uses a multi-scale visualization paradigm, presenting both a large, magnified brain model and a smaller, interactive model. Users interact with the system from a "mission control" platform, using a single VR controller. Three information panels provide supplementary information: a schematic of the anxiety-related functional subsystems, descriptions of brain connectivities, and a progress indicator. The workflow is divided into two phases: anatomical learning and connectivity learning. In the anatomical learning phase, users select key brain structures using the smaller brain model, triggering the display of relevant information. The connectivity learning phase allows users to explore connections between structures and their roles in subsystems. A laser pointer is used to select menu items in the learning material panel. Color-coding consistently links structures, connections, and subsystems. The system was developed using Unity and tested with an HTC VIVE Pro Eye VR headset. A user study with 11 participants assessed the usability of SONIA using the System Usability Scale (SUS), feedback forms evaluating visual design, interaction design, and learning experience, and open-ended questions. Quantitative data was analyzed using one-sample t-tests.

The user study yielded a SUS score of 79.8 ± 11.6, significantly exceeding the usability threshold of 68 (p = 0.007). Positive feedback was observed for overall ease of use, intuitiveness, and consistency. While opinions on system complexity were divided, other subscores indicated positive user interaction. The visual design received an overall score of 3.9 ± 0.5 (p = 0.0001), with high ratings for graphic styles and color-coding. Interaction design effectiveness scored 3.6 ± 0.9 (p = 0.046). Participants strongly felt they had learned a lot (3.9 ± 0.7, p = 0.002), although opinions on the benefit of the multi-scale navigation for anatomical understanding were more neutral (3.1 ± 1.1, p = 0.80). Qualitative feedback revealed positive reactions to the system's utility and novelty, but also suggested that the interface might contain too much visual information, potentially overwhelming some users.

SONIA, as the first VR system integrating descriptive insights into neural pathway exploration and learning module design, effectively leverages VR's advantages in 3D visualization and combines them with carefully designed user interaction strategies. The multi-scale representation, while not decisively proven beneficial for anatomical understanding in this study, contributed to the system's overall visual appeal and enriching learning environment. The positive user feedback on learning outcomes and ease of use supports the effectiveness of the designed interaction paradigms and the incorporation of built-in reward systems, which motivate users to navigate and learn. The system's design accommodates different levels of prior knowledge, making it suitable for users with varying degrees of familiarity with neuroanatomy. Areas for improvement include simplifying the UI and managing information density to minimize potential user overload.

SONIA offers a novel, customizable VR system for learning about functional brain systems and networks. Unlike existing systems that primarily focus on anatomical visualization, SONIA integrates an immersive environment with detailed narratives and user-friendly interaction strategies. User studies confirm its high usability, positive user experience, and educational value. This prototype demonstrates the potential of VR for medical education and exploration and opens avenues for future research and development with diverse learning materials and network models.

The study's sample size was relatively small (11 participants), potentially limiting the generalizability of the findings. The system was tested using only one example of brain pathways (anxiety regulation network) which could limit the findings to just that specific network. Future work should include testing with larger, more diverse populations and evaluating the system's efficacy with additional brain networks and learning materials. Further refinement of the user interface to manage information density will optimize the learning experience.