At-home Pupillometry using Smartphone Facial Identification Cameras

Title of the Paper

At-Home Pupillometry using Smartphone Facial Identification Cameras

Paper Information

• Field of Study: Medical Technology and Smartphone Applications
• Keywords: Smartphone, Facial Recognition, Pupillometry, Portable Medical Devices, Neurological Disease Screening, User Experience, Size Measurement, Remote Health Monitoring

Research Background and Problem Statement

• Identified Issues or Challenges:

  • Existing professional-grade pupillometers are expensive and difficult to widely adopt, limiting their application in telemedicine and at-home neurological disease screening.
  • RGB cameras struggle to accurately measure pupil size, especially under conditions of dark pupils or complex environments.
  • Current research on smartphone-based pupillometry is largely limited to measuring relative size changes, lacking validation for absolute size measurements.
  • High-precision pupil measurement typically requires near-infrared (NIR) cameras and professional equipment, which are rarely available in consumer devices.

• Significance of the Problem:

  • Pupillometry has the potential to detect cognitive load, neurological arousal, and early signs of neurodegenerative diseases (e.g., Alzheimer’s, Parkinson’s, and schizophrenia).
  • Providing a low-cost, widely accessible, and at-home pupil measurement device could advance telemedicine and improve the convenience of early screening.

• Research Motivation and Related Work:

  • The motivation lies in leveraging the ubiquity of smartphones and the latest NIR cameras for pupil measurement to reduce costs and improve accessibility.
  • Limitations of previous studies include low accuracy, lighting constraints of RGB cameras, and the inability to measure absolute pupil size.

Proposed Solution

• Proposed Method or Solution:

  • Utilize the facial recognition functionality in the latest smartphones, combining front-facing NIR cameras with RGB cameras, to develop an attachment-free smartphone application for measuring pupil size.
  • Employ stereo vision to calculate distance, enabling pixel-to-millimeter conversion for precise absolute pupil size measurement.

• Innovations:

  • First validation of smartphones’ ability to measure absolute pupil size, compared against professional clinical pupillometers.
  • Use of NIR cameras to address the low contrast between pupils and irises in RGB imaging, making pupil measurement insensitive to eye color.
  • The application can perform a complete pupil light reflex test without relying on flash, adapting to various home environments.

• Implementation Steps and Key Technologies:

  • Distance Measurement: Calculate the distance between the subject’s eye and the camera using stereo vision differences between the NIR and RGB cameras.
  • Pupil Size Conversion: Convert pixel information in the image to millimeter units to obtain absolute pupil diameter measurements.
  • Data Processing: Use validated neural network models to extract pupil position and size, integrating distance information for final calculations.
  • User Interface Development: Provide an intuitive and user-friendly application to guide users through the test, especially targeting elderly users with limited technical proficiency.

Research Outcomes

• Specific Results:

  • The smartphone system achieved a median absolute measurement error of 0.27 mm compared to professional clinical pupillometers, with a median error of 3.52% for pupil size changes.
  • Two experiments were conducted in home environments, including a light reflex test and a Digit Span Recall Test, progressively validating the device’s accuracy and noise resistance.

• Comparison with Existing Solutions and Advantages:

  • Significantly improved measurement accuracy for dark irises compared to traditional smartphone systems relying on RGB cameras.
  • Operates without a flash, requiring less environmental control than previous systems, making it more adaptable.
  • Outperformed previously reported smartphone systems in the light reflex test.

• Experimental or Evaluation Results:

  • In the light reflex test, the average pupil change range was 2.86 mm, with the smartphone system achieving an average absolute error of 0.39 mm.
  • In the Digit Span Recall Test, the average pupil change range was 1.41 mm, with the average absolute error increasing to 0.96 mm, indicating room for further optimization under complex task conditions.
  • In deployment experiments with elderly users, most of the 15 participants successfully completed the test, though some data were unusable due to low-quality issues (e.g., eyelid occlusion, incorrect operation).

• Limitations and Future Directions:

  • Limitations:
    • The system is sensitive to the accuracy of the distance between the smartphone and the eye, with hand movement potentially affecting measurement precision.
    • Although the user interface was designed for low technical proficiency, some participants still required additional aids (e.g., 3D-printed lens mounts).
    • The system currently relies on specific hardware (e.g., the NIR camera in Google Pixel 4), lacking generalizability to a broader range of devices.
  • Future Directions:
    • Develop more robust distance measurement methods, including adding additional reference points or using iris proportions for further calibration.
    • Optimize user guidance and auxiliary device design to address visual occlusion issues caused by drooping eyelids.
    • Explore broader applications, such as large-scale screening or early intervention tools for neurological diseases.