# Placement-tolerant optics for a smartphone dermatoscope

Canonical: https://www.formlightoptics.com/portfolio/smartphone-dermatoscope-placement-robustness
Markdown: https://www.formlightoptics.com/portfolio/smartphone-dermatoscope-placement-robustness.md

Redesigned a three-lens smartphone clip-on imager so it stays sharp when the phone is not seated perfectly.

Medical device startup (digital dermatology). 2026.

Capabilities: Imaging, Optomechanics, Simulation, Stray light, AI/ML.

### Challenge

The existing three-lens design lost image quality once the phone camera sat a fraction of a millimeter off center, and the outer field collapsed as that offset grew. Real users cannot seat a phone that precisely. Every mechanical interface, from lens thickness to the phone case, had to stay the same.

### Approach

We traced the root cause to the phone aperture acting as a moving system stop, then re-optimized only the molded asphere shapes against a model with the phone deliberately misplaced. A Bayesian optimization campaign scored candidates directly on MTF instead of wavefront proxies. A ghost screen of the internal reflection paths, an AR coating decision, and a full structural-thermal-optical tolerance suite followed.

### Result

Worst-case contrast across a sub-millimeter placement box improved about 100x while the barrel, retention, and case carried over unchanged. The adopted design passed the tolerance load cases.

### Key figures

- ~100x: worst-case contrast under sub-millimeter phone misalignment
- ~300x: lower flare peak at the sensor
- ~15%: light throughput

Tools: Zemax OpticStudio, custom programs and scripts.

Related services: [Imaging lens design](https://www.formlightoptics.com/services/imaging-lens-design), [Zemax OpticStudio consulting](https://www.formlightoptics.com/services/zemax-opticstudio-consulting), [Optomechanical design](https://www.formlightoptics.com/services/optomechanical-design), [Stray light and non-sequential simulation](https://www.formlightoptics.com/services/stray-light-and-non-sequential-simulation).