TL;DR
This paper introduces and evaluates CARDIACAR, a mobile augmented reality (AR) application designed to improve cardiovascular surgical planning. The study's key contribution is the first-ever usability evaluation of a mobile AR surgical planning tool with medical experts, demonstrating its potential to streamline complex procedures and enhance spatial understanding.
What Problem Does This Paper Address?
Planning for complex congenital heart disease (CHD) surgeries traditionally relies on time-consuming and resource-intensive methods like 3D-printing physical heart models. While augmented reality (AR) offers a promising avenue to digitize and streamline this process, there's a significant lack of research evaluating these AR tools with the very medical professionals who would use them. This paper aims to bridge that gap by assessing the usability and technical innovations of a novel mobile AR application, CARDIACAR, with experienced cardiothoracic surgeons and cardiologists.
Methods
The researchers developed CARDIACAR, an iOS application leveraging ARKit for mobile AR experiences. The application offers several interactive features for surgical planning:
- Importing 3D heart models: Users can import patient-specific 3D heart models.
- Model Viewing: Standard AR gestures allow for rotation and resizing of the model in a real-world environment.
- Omni-directional Slicing: This is a novel feature that allows users to preview and visualize cross-sections of the 3D heart model in real-time. The system uses a custom surface shader to highlight the planar cross-section, overcoming limitations of standard ARKit/SceneKit for hollow geometries.
- Virtual Annotation: Users can add notes and labels directly onto the 3D model within the AR environment.
A usability study was conducted with six medical experts (four cardiothoracic surgeons, two cardiologists) from Children's Healthcare of Atlanta (CHOA). Participants were given a brief tutorial and then asked to use CARDIACAR's features, followed by a period of free exploration. Feedback was collected through Likert scale ratings for usability and usefulness, as well as qualitative comments on how the features could aid surgical planning and potential improvements.
Datasets
The study utilized a 3D heart model provided by Children's Healthcare of Atlanta (CHOA), which was constructed from a patient's de-identified medical imaging data. No new datasets were introduced or benchmarked in this paper.
Key Insights and Ideas
- Mobile AR is a viable tool for surgical planning: The study demonstrates that mobile AR, accessible via common devices like iPhones and iPads, can be effectively used for complex surgical planning tasks.
- Innovative slicing is crucial: The custom omni-directional slicing feature was a significant technical innovation, enabling detailed visualization of internal cardiac structures that is difficult with traditional methods or simpler AR slicing implementations.
- Portability enhances spatial understanding: The mobile nature of the tool allows surgeons to physically move around the AR model, gaining a more intuitive "inside" view of the heart, which is a key advantage over static 3D prints.
- User feedback drives practical improvements: The direct input from medical experts highlighted specific areas for refinement, such as gesture sensitivity and the potential for dedicated annotation modes, showing the value of expert-driven design.
- Different user needs for AR immersion: A notable observation was the differing preferences between surgeons (preferring a model-only view) and cardiologists (benefiting from seeing the physical background for spatial anchoring), suggesting that AR immersion levels might need to be customizable.
Key Results
- High Usability and Usefulness Ratings: Participants generally rated the features highly. Omni-directional slicing received an average usability rating of 4.5. Model viewing and annotation features were also rated very positively (4 and 5 respectively).
- Qualitative Support for AR: Medical experts strongly agreed that CARDIACAR's features, particularly the slicing and portability, would benefit real-life surgical planning by providing a better understanding of cardiac geometry than traditional 3D-printed models.
- Annotation for Practical Scenarios: The annotation feature was deemed highly helpful for labeling important regions, with cardiologists suggesting it could be a dedicated mode.
- Efficient Model Import: The model import feature was rated highly (4.5 usability) for its ability to facilitate patient-specific analysis and efficient exploration of various heart models.
Limitations and Future Work
- Limited Sample Size: The study involved only six medical experts, which is a small sample size.
- Specific Use Case: The evaluation focused on congenital heart disease planning.
- Gesture Sensitivity: Surgeons noted that the gesture sensitivity for model manipulation was initially too high, leading to rapid movements.
- Future Work: The authors suggest refining gesture movements and the user interface for slicing and panning. They also propose a larger-scale evaluation with more participants and a wider variety of surgical procedures to further assess CARDIACAR's usability in different contexts. Further development could also explore customizable AR immersion levels based on user preferences.
Summary updated