Online view enhancement for exploration inside medical volumetric data using virtual reality
Hongkun Zhang,Lifeng Zhu,Qingxiang Zhang,Yunhai Wang,Aiguo Song
DOI: https://doi.org/10.1016/j.compbiomed.2023.107217
Abstract:Background and objective: Medical image visualization is an essential tool for conveying anatomical information. Ray-casting-based volume rendering is commonly used for generating visualizations of raw medical images. However, exposing a target area inside the skin often requires manual tuning of transfer functions or segmentation of original images, as preset parameters in volume rendering may not work well for arbitrary scanned data. This process is tedious and unnatural. To address this issue, we propose a volume visualization system that enhances the view inside the skin, enabling flexible exploration of medical volumetric data using virtual reality. Methods: In our proposed system, we design a virtual reality interface that allows users to walk inside the data. We introduce a view-dependent occlusion weakening method based on geodesic distance transform to support this interaction. By combining these methods, we develop a virtual reality system with intuitive interactions, facilitating online view enhancement for medical data exploration and annotation inside the volume. Results: Our rendering results demonstrate that the proposed occlusion weakening method effectively weakens obstacles while preserving the target area. Furthermore, comparative analysis with other alternative solutions highlights the advantages of our method in virtual reality. We conducted user studies to evaluate our system, including area annotation and line drawing tasks. The results showed that our method with enhanced views achieved 47.73% and 35.29% higher accuracy compared to the group with traditional volume rendering. Additionally, subjective feedback from medical experts further supported the effectiveness of the designed interactions in virtual reality. Conclusions: We successfully address the occlusion problems in the exploration of medical volumetric data within a virtual reality environment. Our system allows for flexible integration of scanned medical volumes without requiring extensive manual preprocessing. The results of our user studies demonstrate the feasibility and effectiveness of walk-in interaction for medical data exploration.