Project: Evaluating Uncertainty Visualization through a game in 3D slicer

[mahsageshvadi]

Draft Status

Ready - team will start page creating immediately

Category

IGT and Training

Presenter Location

In-person

Key Investigators

• Mahsa Geshvadi ( Brigham and Women's Hospital, USA)
• Sarah Frisken ( Brigham and Women's Hospital, USA)

Project Description

Uncertainty is present in all medical images and originates from different sources. Uncertainty is difficult to interpret due to its probabilistic nature, and communicating it is equally difficult. In this project, we developed an uncertainty visualization module on 3D Slicer, enabling users to explore different uncertainty visualization techniques. The key challenge now is evaluating these techniques for real-world applicability. To address this, we implemented a game for quantitative evaluation of uncertainty visualization. In the game, users perform tasks requiring decision-making under uncertainty, and we measure their performance with and without visualization using scores. Specifically, the game simulates decision-making during tumor resection surgery, where MRI images have uncertainty due to brain shifts. Users must decide whether to carve out the tumor at specific locations, reflecting the real surgical decision-making process.

Objective

1. Objective A: Uncertainty visualization helps make better decisions under uncertainty.
2. Objective B: Exploring and evaluating the Uncertainty Visualization module with a game helps improve the understanding of uncertainty.

Approach and Plan

1. Training level: helps participants become familiar with uncertainty in the game and explore different visualization techniques. Players can see the ground truth segmentation and their scores, allowing them to observe the impact of their decisions directly.

Training.mov

3. The challenge phase builds on skills learned during training, where players perform tasks without seeing the ground truth segmentation or their scores and cannot undo actions. It consists of two steps: first, making decisions without uncertainty visualizations on a different case, and second, repeating the task with uncertainty visualizations on the same case, without any prior feedback or guidance.

Challenge.mov

Progress and Next Steps

1. Explore uncertainty visualization evaluation through a game in more scenarios.
2. Make the game more complex by adding new levels.

Illustrations

No response

Background and References

https://github.com/mahsageshvadi/UncertaintyVisualization

[github-actions]

Project Page Pull Request Creation

✅ COMPLETED: See #1195