Smartphone Tear Film Imaging for Dry Eye Testing

ISEF Category: Biomedical Engineering

Subcategory: Biomedical Sensors and Imaging  ·  Difficulty: Advanced  ·  Setup: School Lab  ·  Time: 1 to 2 Months

The Hook

Dry eye affects millions of people, and the first signs often show up before someone feels real discomfort. Your phone camera can help catch those signs if you can measure the tear meniscus, the thin liquid edge along the eyelid. With the right optics and image analysis, a cheap setup can act like a tiny diagnostic tool. That makes this a strong science fair project with real medical relevance.

What Is It?

This project looks at the tear meniscus, the small curved strip of fluid that sits along the lower eyelid. Think of it like the waterline in a glass. When your tear film is healthy, that waterline stays stable. When the eye dries out, the line gets thinner, breaks sooner, and changes shape.

You can capture that line with a smartphone, a clip-on macro lens, and a slit light. The slit light gives the tear edge a bright, narrow highlight, which makes it easier to detect in an image. Then software can measure meniscus height, or the thickness of that tear band, frame by frame. The break-up time part tells you how long the tear film stays smooth before it starts to fragment. That gives you a number you can compare across people, conditions, or time.

Why This Is a Good Topic

This is a strong science fair topic because you can measure a real biomedical signal with low-cost hardware and clear image data. You can test how lighting, blinking, screen use, or eye drops affect tear-film stability. The project connects to dry eye, a common health issue, and teaches optics, segmentation, calibration, and regression. You can also make it more original by comparing your measurement method against a simpler visual score or a different imaging setup.

Research Questions

• How does slit-light angle affect the accuracy of tear-meniscus height measurement?
• What is the effect of clip-on macro-lens magnification on segmentation quality for the tear meniscus?
• Does automated meniscus-height regression agree with manual measurement across different image sets?
• To what extent does screen exposure before imaging change tear-film break-up time?
• Which smartphone camera settings produce the most stable tear-edge detection?
• How does the presence of eyelash occlusion affect segmentation error in tear-meniscus images?

Basic Materials

• Smartphone with a good rear camera.
• Clip-on macro lens.
• Small LED slit light or narrow-beam LED source.
• Matte black background or cardboard hood.
• Tripod or phone stand.
• Ruler or printed calibration target.
• Image analysis software with measurement tools.
• Notebook or spreadsheet for data logging.

Advanced Materials

• Smartphone with manual camera controls.
• Clip-on macro lens.
• Adjustable slit-lamp style LED source.
• Chin rest or head stabilization setup.
• Calibration phantom or scale target.
• Computer for segmentation and regression analysis.
• MATLAB, Python, or equivalent image analysis environment.
• Annotation tool for training or checking segmentation outputs.
• Reference imaging setup for repeatability testing.

Software & Tools

• ImageJ: Measures tear-meniscus height, compares frames, and checks calibration.
• Python: Automates image processing, regression, and error analysis.
• Jupyter Notebook: Keeps code, plots, and notes in one place.
• SAM2: Segments the tear meniscus and helps separate it from the eyelid and background.
• Google Sheets: Organizes measurements, image labels, and trial metadata.

Experiment Steps

1. Define the exact signal you will measure, such as tear-meniscus height, break-up time, or both.
2. Choose one imaging geometry first, then hold the camera, light, and subject position constant.
3. Build a calibration plan so pixel measurements become real-world units.
4. Decide how you will separate the tear edge from eyelid tissue, lashes, and reflections.
5. Create a comparison method between automated segmentation and manual scoring.
6. Plan repeat trials across subjects or conditions so you can test consistency and error.

Common Pitfalls

• Using room lighting that changes from trial to trial, which shifts contrast and breaks image thresholding.
• Letting the phone move between frames, which makes meniscus height look like it is changing when it is not.
• Ignoring calibration scale, which leaves results in pixels instead of a usable unit.
• Confusing tear reflections with the real tear edge, which causes the segmentation model to trace the wrong boundary.
• Mixing up blink timing and image timing, which makes break-up time measurements hard to compare.

What Makes This Competitive

A strong version of this project does more than make a phone picture look better. You need a clear measurement model, a careful calibration method, and a direct comparison against human scoring or another imaging approach. Strong entries also test error sources, like lighting angle, eyelash occlusion, or camera distance, then show how much each one changes the final result. That kind of analysis turns a neat demo into real engineering.

Project Variations

• Test how different slit-light colors affect tear-edge contrast and segmentation accuracy.
• Compare healthy and dry-eye style imaging conditions by measuring how blink timing changes tear-film break-up patterns.
• Use multiple phone models to see whether sensor differences change meniscus-height regression.

Learn More

• PubMed: Search for review articles on tear meniscus height, tear-film break-up time, and dry eye imaging.
• NIH National Eye Institute: Read patient and research pages on dry eye and tear-film basics.
• ImageJ Documentation: Find guides for image calibration, edge measurement, and frame analysis.
• MIT OpenCourseWare, Biomedical Optics: Search the course site for optics, imaging, and biomedical sensing lectures.
• Journal of Biophotonics: Search for peer-reviewed studies on ocular imaging and tear-film measurement.