Investigating Clutching Interactions for Touchless Medical Imaging Systems

Document Title

Investigating Clutching Interactions for Touchless Medical Imaging Systems

Document Information

• Subject Area: Human-Computer Interaction (HCI), Medical Imaging Systems
• Keywords: Clutching, Gestures, Medical Imaging Systems, Midas Touch, PACS, Touchless Interaction

Research Background and Problems

• Identified Issues or Challenges:
  • Traditional input devices like mice and keyboards may prevent medical personnel from directly operating medical imaging systems in sterile environments.
  • Unintended inputs in touchless interaction (e.g., misrecognized gestures or voice commands) are particularly disruptive in medical scenarios, potentially interrupting workflows or even affecting patient treatment.
• Significance:
  • The strict sterility requirements in medical environments necessitate minimizing contact with shared device surfaces while ensuring reliable interaction methods.
  • Medical imaging systems (e.g., PACS) are critical in clinical workflows, as rapid access to and manipulation of patient images can improve the accuracy and efficiency of treatment.
• Research Motivation and Related Work:
  • While current touchless technologies have garnered attention, there is a lack of in-depth research on effectively distinguishing intentional inputs from non-interactive actions.
  • Previous studies have proposed some touchless interaction solutions, but their broad practical application in medical environments faces many challenges.

Solution

• Proposed Methods or Solutions:
  • A prototype touchless PACS interface was developed, integrating four "Clutching" mechanisms (for switching input modes):
    1. Gestures
    2. Voice
    3. Active Zone
    4. Gaze
  • These methods are used to activate or deactivate system input, reducing misrecognition and improving reliability in medical environments.
• Innovations:
  • For the first time, a comprehensive comparison of the performance of four clutching mechanisms (explicit and implicit methods) was conducted, revealing their applicability in medical use from a user experience perspective.
  • Investigated how to design and integrate touchless interaction in real hospital scenarios.
• Implementation Steps and Key Technologies:
  • Experiment design: Collaborated with 34 medical professionals to simulate intraoperative and postoperative scenarios.
  • Defined experimental tasks: Included common PACS operations such as browsing patient images, manipulating images (e.g., zooming, rotating), and viewing reports.
  • Built an interactive system using Azure Kinect and related cloud services for voice recognition, gaze detection, and gesture tracking.

Research Findings

• Specific Results:
  • Quantitative Results: The Active Zone method demonstrated the best performance in terms of switching success rate (100%) and overall user task load (NASA-TLX scale scores).
  • Qualitative Feedback:
    • Advantages and disadvantages of different clutching methods:
      • Gesture Method: Easy to learn but affected by protective equipment.
      • Voice Method: Intuitive to use but prone to voice recognition failures and environmental noise interference.
      • Gaze Method: Highly automated but prone to misinterpretation, leading to a lack of user control.
      • Active Zone Method: Seamlessly integrates into user operation habits and is highly compatible with existing workflows.
    • Medical professionals emphasized that directly controlling PACS without relying on intermediaries significantly improves work efficiency.
  • Highlighted how touchless interaction in medical environments can optimize user experience by reducing pathogen transmission through surface contact.
• Advantages Over Existing Solutions:
  • The Active Zone method naturally integrates into existing clinical workflows and imposes lower cognitive load compared to other explicit methods.
  • Proposed a secure authentication scheme for touchless interaction, discussing the potential to enhance safety and efficiency.
• Experiments and Evaluation:
  • A diverse set of experimental tasks, method comparisons, and analyses provided a systematic evaluation.
  • The experiments yielded critical quantitative data (e.g., switching success rates, operation times) and in-depth user feedback.
• Limitations and Future Directions:
  • Limitations:
    • The controlled experimental environment may not fully reflect real hospital scenarios (e.g., noise, crowding).
    • Current methods do not address deployment challenges in multi-user or dynamic situations.
  • Future Directions:
    • Explore the dynamic configurability of Active Zones and mechanisms for identifying the "dominant user" in multi-user environments.
    • Integrate multimodal touchless interaction methods (e.g., combining voice, gestures, and gaze) to further optimize user experience.
    • Address challenges such as environmental noise interference in voice recognition and PPE detection accuracy in medical settings.