MRD1 mm
Margin–reflex distance 1
Corneal centre up to the upper lid margin, on the same vertical.
An iPhone and Mac app for oculofacial plastic surgery research.
Capture and organise your study photographs, record depth with LiDAR, and segment faces with built-in AI models. OphthoRuler then measures MRD1, MRD2, and other common values automatically, and analyses them with comprehensive statistics for common oculoplastic research designs.
R Right eye 0.0566 mm/px
- MRD1
- 2.8 ±0.4 mm
- MRD2
- 5.2 ±0.4 mm
- PFH
- 8.0 ±0.5 mm
- WTW
- 11.71 ref mm
01 · Measure
Twelve measurements built in, and a visual editor for your own.
The bundled Oculoplastic core set covers the standard periocular distances, plus palpebral fissure area, lid heights, and white-to-white. Each is built from labelled landmarks, so you can read it, change it, or build your own.
-
-
MRD2 mm
Margin–reflex distance 2
Corneal centre down to the lower lid margin.
-
PFH mm
Palpebral fissure height
Upper to lower lid margin through the corneal centre.
-
PFW mm
Palpebral fissure width
Medial to lateral canthus of the same eye.
-
ICD mm
Intercanthal distance
Medial canthus to medial canthus.
-
IPD mm
Interpupillary distance
Corneal centre to corneal centre.
-
OCD mm
Outer canthal distance
Lateral canthus to lateral canthus.
-
Tilt deg
Canthal tilt
Angle of the canthal axis against the horizontal.
-
+4
PFA · Upper lid height · Lower lid height · WTW
MRD1, as the app defines it
Measurements are defined by editable formulas.
Every measurement is two landmarks and a method. MRD1, for example, is the vertical distance from the cornea centre to the highest point of the upper lid margin directly above it. Switch from the Visual Formula Editor to the Text Formula Editor to see the same definition written as a formula. You can duplicate or edit any measurement, or share a whole set with a collaborator.
02 · On the Mac
From photograph to figure, in one place.
-
01 Design
Plan the study before you collect a single photo.
A step-by-step wizard helps you set up your study: pick a design, decide how to handle both eyes, plan your visits and photo series, and define your variables. Your plan is saved as version 1, and any later change is saved as a new version, so you can always see what changed.
-
02 Categorise
Organise photographs by participant and visit.
Group photographs into participants, encounters, and photo series, tag them, and run a model across a selection. Photographs stay in the encrypted library on your Mac.
-
03 Annotate
Landmarks are placed for you, and you can edit any of them.
On-device models trace lid margins, brows, iris boundaries, and canthi, with the right eye in red and the left in blue. Nudge a point, redraw a curve, or run a newer model: manual edits are never overwritten.
-
04 Measure
Click a value to see how it was measured.
Choose MRD1 and the view zooms to the eye, marks both points, and shows the definition beside the number and its ± uncertainty.
-
05 Export
Figures ready for a manuscript.
Export a figure with the annotated photograph, cropped views of each eye, colour-coded right and left eyes, and a table of measurements with their uncertainties. You choose where the files are saved.
-
06 Compose
Or build your own figure.
Pick photographs, choose the crop and layout, then switch on the measurements to draw as dimension lines. Labels move where you drag them.
-
07 Data
One table per study.
Measured columns arrive locked and labelled with their source. Add your own study variables, paste from a spreadsheet, or export CSV for R.
03 · Calibration
Every measurement records how it was calibrated.
A measurement is only as good as its scale, so every value keeps a record of how it was calibrated. On an iPhone with LiDAR or TrueDepth sensors, the scale is calculated automatically. For photos from other cameras, or from devices without a depth sensor, you can still use the cornea white-to-white or an object of known size as a reference.
- 01
Default with depth
LiDAR and TrueDepth
The rear LiDAR scanner or the front TrueDepth camera saves a depth map with every photo. Face distance and focal length give a measured scale, with no sticker or ruler in the frame.
- 02
Front camera
Depth (3D)
ARKit’s depth-tracked face mesh measures the face in three dimensions at capture, so the scale comes from the camera itself rather than an assumption.
- 03
Any photo
White-to-white
The iris width is taken as the corneal diameter: 11.71 mm by default, or a value you have measured for that eye.
- 04
In the frame
Size reference
Draw a rectangle over an object of known size in the photo and enter its dimensions.
- Post-it 76 × 76 mm
- 10 mm sticker
- Coin
- Ruler
- 05
Known value
Manual scale
Enter millimetres per pixel directly.
04 · Models
Choose the model that suits your study.
OphthoRuler is model-agnostic. Each segmentation model is a card in the Model tab, listing the landmarks it produces, the measurements that come with it, its authors, licence, and citation. Measurements reference landmark labels rather than a particular network, so the same MRD1 works whichever model placed the points.
- Built-in
FaceMesh
Face and iris landmarks from a photo
- Built-in
LightIrisNet Pupillometer
Pupil and iris boundaries from a visible-light photo
- Built-in
GazeTracking Motility
Pupil position per gaze for the 9-gaze motility chart
- Built-in
Periocular-Vision
Face landmarks from Apple’s Vision framework (fallback)
- Soon
Add your own model
Package a Core ML model with its label map, licence, and citation. Share it so that other research groups can use (and cite) your model with their own studies.
-
Measurements follow labels
Measurement sets refer to landmark labels rather than to a model. A set marked Any model works with whichever model produced those labels.
-
Authors credited on every card
Authors, licence, model page, source repository, and paper sit beside each model, with the citation ready to copy.
-
Runs on your device
Every model runs through Core ML on your Mac or iPhone. Photographs are never uploaded to be analysed.
05 · On the iPhone
Capture on the iPhone, check on the spot.
On-screen guides frame the face or a single eye, photo series step through each gaze, and the landmark model runs on the phone, so you can check a capture before the participant leaves.
- A
Quick measurements
Oculoplastics, levator function, pupillometer, and a 9-gaze motility panel, each with its own guide.
- B
Guided photo series
Primary, up-, down-, and side gaze in order, with retakes one tap away.
- C
Pupillometry in light and dark
Flash-lit pairs, then each eye again under an infrared flashlight for dark diameters.
- D
Edit by touch
Drag a lid point or resize an iris and watch MRD1 or WTW update as you go.
06 · Analyse
Statistics that account for two eyes per participant.
Describe your study in plain language. OphthoRuler chooses a suitable model, accounts for the fellow eye, draws the figures, and drafts the methods text with every number linked to its result.
-
01 Design
Start from your study design.
Choose from ten study designs, each described by the question it answers. The model, figures, tables, and manuscript wording follow from that choice.
-
02 Eyes
Decide how the two eyes are counted.
Choose how each participant’s eyes enter the analysis: both eyes in a clustered model, one prespecified eye, the affected eye, or the mean of the two. Treating every eye as an independent patient is not offered.
-
03 Endpoint
Lock your primary endpoint in advance.
Pick the measurement, the timepoint, and the comparison for your primary endpoint, then lock it. Secondary endpoints use the same model, with optional Holm or Benjamini–Hochberg adjustment. Changing a locked plan creates a new, recorded version.
-
04 Results
Results with every participant plotted.
The headline result is an effect estimate with its 95% confidence interval, shown alongside the model and the sample sizes. Figures plot each participant, not only the mean.
-
05 Figures
Adjust figures to fit the journal.
Edit titles, axis labels, point size, and palette, including the colour-blind-safe Okabe–Ito palette. Figures export as vector SVG and 300 dpi PNG.
-
06 Guidance
Explanations written for your data.
Warnings and notes are specific to your data. For example, canthal tilt is an angle, so it is summarised on a circle with a mean direction and a circular SD, and the app explains why.
-
07 Tables
Tables that count participants and eyes separately.
Table 1 reports participants and eyes separately. Table 2 lists every endpoint with its estimate, interval, p-value, and model.
-
08 Manuscript
Methods and results drafted for you.
The Methods and Results are automatically drafted for you based on the selected study design, analyses, and generated results. The app lets you export everything as a report bundle with documents, figures, and tables included.
-
09 Guide
A built-in statistics guide.
Plain-language articles cover every model, the two-eye problem, circular statistics, reliability, diagnostic accuracy, and multiple comparisons, one click from any result.
Examples of supported study designs
Each design ships with a Sample Study of synthetic data, so you can explore the study design before collecting any real data.
-
Before vs after
Did MRD1 change after ptosis repair?
Δ 2.19 mm 95% CI 1.99 to 2.38
Enlarge the Before vs after figure -
Before vs after, two groups
Did MRD1 change more after surgery than with observation?
Δ 2.13 mm 95% CI 1.87 to 2.40
Enlarge the Before vs after, two groups figure -
Longitudinal
How does brow height settle after a brow lift?
Δ 2.36 mm Month 12 · 95% CI 2.01 to 2.72
Enlarge the Longitudinal figure -
Two groups
Which lifts the lid more: levator advancement or Müller resection?
−0.55 mm 95% CI −0.82 to −0.29
Enlarge the Two groups figure -
Three or more groups
Do three ptosis techniques give different results?
F 4.76 Global test · p = 0.01
Enlarge the Three or more groups figure -
Fellow-eye
Does a new crease fixation differ from the standard one in the fellow eye?
0.43 mm 95% CI 0.23 to 0.62
Enlarge the Fellow-eye figure -
Normative
What is a normal palpebral fissure height?
PFH 9.82 mm Reference range 8.59 to 11.05
Enlarge the Normative figure -
Reliability
Do manual and automated MRD1 measurements agree?
ICC 0.90 95% CI 0.77 to 0.95
Enlarge the Reliability figure -
Diagnostic
How well does an AI score match surgeon grading of ptosis?
AUC 0.97 95% CI 0.94 to 1.00
Enlarge the Diagnostic figure -
Prognostic
Does levator function predict success?
OR 0.51 Recurrence · 95% CI 0.37–0.70
Enlarge the Prognostic figure
-
Built around your design
Pick one of ten study designs, from before-and-after to diagnostic validation. The analysis follows from your question.
-
Estimates with intervals
Every result is an effect estimate with a 95% confidence interval, not only a p-value.
-
Plain-language warnings
The app flags small samples, unlocked endpoints, and data leakage, and says why.
-
Checked against R
Mixed models, ICC, Bland–Altman, ROC, and risk models are cross-validated against reference R implementations.
07 · Privacy
Your data stays on your devices.
There is no OphthoRuler server. The developers never see your photographs, participants, or numbers.
-
On-device models
AI models are run locally on your Mac or iPhone using Core ML. No image is uploaded to the internet to be analysed.
-
End-to-end encrypted
Photographs and study records are encrypted with AES-256-GCM using a password only you know.
-
No internet connection needed
All photos, AI models, and statistical analyses are carried out locally on your iPhone and Mac.
-
Local-network sync
Mac and iPhone pair by QR code and sync over your own network, encrypted in transit. No cloud.
-
Metadata stripped
EXIF, GPS, and maker notes are removed on capture and import. A face is still identifiable, so this reduces risk but does not de-identify a photo.
-
No accounts
You own your data. Your data and photos are never uploaded anywhere by OphthoRuler. All data is stored locally on your iPhone and Mac by default.
Research use only
OphthoRuler is Research Software.
OphthoRuler is experimental research software. It has not been submitted for medical-device authorisation, is not certified against any privacy or health-data standard, and can be wrong.
- Do not use it to diagnose, screen, monitor, treat, or plan care for anyone.
- Check every value it produces before you rely on it or publish it.
- Ethics approval, consent, and data protection for your study remain yours to obtain.
06 · Questions
Common questions.
Who is OphthoRuler for?
Researchers in oculoplastics, ophthalmology, and facial anthropometry who photograph eyes and need consistent, reproducible periocular measurements and a sound way to analyse them.
What do I need to run it?
A Mac with macOS 15 or later for annotation, measurement, export, and analysis, and optionally an iPhone with iOS 18 or later for guided capture. The two sync directly over your local network.
How accurate are the measurements?
The app reports every value with a propagated ± uncertainty and records how it was calibrated. Accuracy depends on calibration, head pose, lens and focus, and the landmarks themselves. It also depends on the quality of the segmentation, whether from the AI model you chose or from landmarks you edited by hand. Check the output, and edit landmarks where the model is wrong. As new and improved AI models are developed, they can be installed and used within OphthoRuler.
Is OphthoRuler for clinical use?
No. OphthoRuler is not a medical device and must not be used to diagnose, screen, monitor, or plan treatment. We built OphthoRuler to facilitate research studies in oculoplastic surgery. It is your responsibility to obtain REB/IRB approval to use OphthoRuler in your studies.
Where is my data stored?
Only on your own devices, encrypted with your password. There is no cloud service and no developer access. Exports go only where you save them. This also means that if you lose your password we are unable to help you recover your data. Make sure you keep multiple backups in various formats and locations.
Can I add my own measurements or models?
Yes. Measurements are built from landmarks in a visual editor (with a text view of the underlying formula) that you can inspect, duplicate, and edit, and measurement sets can be imported and exported. Bundled models ship with their licences and citations; adding and sharing your own models is coming during the beta.
How do I get it?
OphthoRuler is in beta. Join the waitlist below and tell us a little about your research. We will be in touch as places open.
Try it on your next study.
Beta · macOS 15+ and iOS 18+
Join the beta waitlist
Tell us a little about your research. We open places in small groups and will email you when yours is ready.