I’ve worked in medical device quality for seven years, four of them at Topcon. I review imaging systems and diagnostic platforms before they ship, and I sign off on hundreds of units a year. In Q1 2024 alone, I rejected roughly 6% of first deliveries because something didn’t survive the validation protocol—calibration drift, firmware oddities, documentation gaps. That’s my job.

People assume that makes me obsessed with specs. It doesn’t. It makes me obsessed with certainty. And that’s the right lens for a question I hear constantly: how does OCT imaging work?

You can describe it as an optics question. I’m going to explain that part. Then I’m going to explain why optics alone won’t help you choose a good device.

How Does OCT Imaging Work? The Optical Answer

OCT—optical coherence tomography—is often compared to ultrasound, except it uses light instead of sound.

In an OCT system, a low-coherence light source (usually a superluminescent diode) is split into two paths by an interferometer. One path, the reference arm, sends light to a mirror at a known distance. The other path, the sample arm, scans the tissue you want to image. When light reflects back from different depths in the tissue, it carries a tiny delay relative to the reference reflection.

You can’t measure that delay directly. Light travels too fast. Instead, the device measures the interference pattern between the two returning beams. Interference only happens when the two path lengths match within the coherence length of the light source. By moving the reference mirror, or by analyzing the spectrum of the combined signal mathematically, the instrument reconstructs depth information for each point. Scan adjacent points, and you get a cross-section: a tomogram.

Modern spectral-domain OCT does this fast enough to produce clear, high-resolution cross-sections of the retina. The layers you see in a clinical scan aren’t reconstructed by guessing—they’re measured.

The resolution is striking. A widely used technical reference for OCT resolution is:

Δz = (2 ln 2 / π) × (λ₀² / Δλ), where λ₀ is the central wavelength and Δλ is the bandwidth of the light source.

At typical retinal imaging wavelengths, axial resolution is around 5–10 micrometers. To put that in context, a red blood cell is about 8 micrometers wide. You’re measuring structures smaller than a single cell.

That’s why engineering details matter so much. An OCT system is an interferometer: its entire measurement depends on aligning two light paths to sub-micrometer precision. Every lens mount, every fiber connector, and every coefficient of thermal expansion inside the chassis affects the result.

Specifications Describe a Device. They Don’t Guarantee It.

A spec sheet tells you what a device should do. It doesn’t tell you whether this particular unit still does it after three hours of continuous use, at 31°C ambient temperature, or after the scanner was jostled during shipping.

In our Q1 2024 quality audit, we flagged a batch of optical engine modules from an outside supplier. Their test reports showed every unit passing. When we ran our own extended validation—warming the units, cycling power, scanning a test target repeatedly—roughly one in ten drifted outside our baseline.

If I had accepted the supplier’s test report at face value, we would have shipped systems that worked in a demo but degraded unpredictably. We rejected the batch. Their first response was that our protocol exceeded industry standards. Maybe it did. But “industry standard” is a floor, not a goal. The question quality work asks isn’t “does it pass minimum?” It’s “will it hold up at the moment a patient is waiting and the clinician has one chance to get the image right?”

For an OCT system, the difference between a stable optical path and one that slowly drifts can be the difference between a reliable screening tool and a machine that produces images nobody fully trusts.

What a Misunderstood Deadline Taught Me

I didn’t always appreciate certainty. I learned it through a communication failure.

This was back in 2021, when I was working with an outside calibration laboratory. I told them: “We need these reports by the 14th.” They heard “sometime that week if it’s convenient.” I found out on the afternoon of the 14th when I called and the reports weren’t ready. Our entire release schedule slipped eleven business days while they reissued documents with the correct reference standard.

We had been using the same words and meaning different things. “By the 14th” sounded unambiguous to me. To them, it was a rough target.

Now every purchase order I write includes explicit lead-time language, and every vendor knows I will enforce it. Some people read that as administrative stubbornness. It isn’t. In medical device production, a missed date doesn’t just sit in a spreadsheet. It means a clinic that planned its patient schedule around the device has to cancel, reschedule, and explain to people why they’re waiting.

The Expensive Kind of Cheap

Let me be direct: the most expensive piece of diagnostic equipment isn’t the one with the highest price tag. It’s the one that fails at the wrong time.

A hospital we work with once bought a refurbished OCT system to save around $14,000 versus buying through an authorized channel. For eight months, it worked well. Then a component in the optical path started causing artifacts that looked like retinal thickening on some scans. The refurbisher had to ship components back and forth for diagnosis, which took two weeks. The replacement part ultimately cost $9,500. Total downtime: about three weeks.

During those three weeks, the system lost more than $18,000 in imaging revenue. The $14,000 “saving” became a $23,000 loss, not counting patients who had to reschedule.

That math isn’t unique to OCT. A spirometer with a small flow-calibration error can produce pulmonary function data that looks plausible but isn’t reliable—and the cost of that error shows up when a clinician makes a treatment decision from bad numbers. An endoscope with inconsistent white balance can make subtle mucosal changes look unremarkable. Different devices, same risk: a small uncertainty in the measurement becomes a large uncertainty in patient care.

Software flags don’t solve this. Software only checks what someone programmed it to check.

Deadlines Make Certainty a Clinical Issue

There’s another side to certainty that has nothing to do with the device itself: time.

If a clinic has one OCT device and it goes down, downtime isn’t measured in hours. It’s measured in canceled appointments, extended waitlists, and referral relationships that weaken when patients are sent elsewhere. That’s why response time and parts availability deserve as much attention as image quality.

Buying from a company with genuine local infrastructure changes the calculation. Topcon doesn’t just sell imaging systems; the network of Topcon Solutions store locations across the country functions as a service logistics network. Spare parts are stocked in places a technician can reach in hours instead of in a warehouse on another continent.

I’m not saying that because it sounds nice. I’ve seen the alternative. When you order a replacement module from a distant supplier with no local distribution, each shipping cycle costs you 11 to 14 days. In a year, those delays add up to more downtime than the price difference between that supplier and a local, authorized partner.

Topcon Electronics is part of the same quality conversation. The embedded electronic modules I review for imaging systems get tested beyond their nominal range—not just checked for “does it switch on.” Temperature coefficients, power stability, and electromagnetic interference are all measured before a module is approved. That’s the kind of engineering margin that never appears on a spec sheet but shows up in the device’s behavior over time.

When your project has a fixed date—an accreditation deadline, a grant-funded screening program, or a visiting specialist clinic—uncertain delivery and uncertain repair times are not neutral facts. They’re risks. And risk is expensive.

The Fair Objection: Aren’t You Just Saying Buy Premium?

It’s a fair question. My answer: no.

I’ve chosen lower-cost components when the application didn’t demand extra certainty. Last year, we compared a generic endoscope light source against the original equipment manufacturer’s version. Generic cost 35% less and the specifications looked identical. But the generic supplier couldn’t commit to a delivery date for their next production run. The OEM version could be at our dock in four days.

The project had a fixed launch date tied to a state screening grant. A delay of three weeks would have cost more than the price difference. We paid more and shipped on time.

That’s the core of the time-certainty argument: when you’re facing a deadline, “probably fine” is the most expensive option in the room. You’re not paying extra for speed. You’re paying to remove the possibility of a delay. In many cases, that removal is worth more than the device itself.

Another objection I hear is “we can test the equipment when it arrives and reject it if there’s a problem.” Testing incoming equipment requires calibrated references, trained personnel, and enough volume to make the process meaningful. Most clinics have none of those. That’s not an insult—it’s reality. It’s why you buy from an organization that does the verification before the device arrives, not after.

My Final Answer

So, how does OCT imaging work? It uses low-coherence light and interferometry to measure microscopic differences in the time it takes light to return from different layers of tissue. That’s the optical story, and it’s a remarkable one.

The practical story is different. The real question isn’t just how OCT works. It’s whether the device in front of you will work the same way next Tuesday, after a power surge, in a warm equipment room, with a full schedule of patients waiting.

I’ve rejected about 6% of first deliveries this year not because those systems would fail every patient. I rejected them because they might fail someone—and in a diagnostic setting, “might” is not a quality standard.

When you compare imaging systems, compare more than hardware prices. Add the cost of a delayed delivery, a long wait for service, an unreliable calibration, and the clinical decisions made from images you can’t fully trust. Then ask yourself what certainty is worth.

In my experience, it’s worth every dollar of the premium.