Transducin: an open-source pipeline recovering SNOMED-CT coded measurements from the undocumented Optopol .OPT and Zeiss Cirrus private-tag formats as DICOM Structured Reports
This paper introduces Transducin, an open-source Python library that reverse-engineers undocumented Optopol .OPT files and Zeiss Cirrus private DICOM tags to generate standardized SNOMED-CT coded DICOM Structured Reports, achieving 100% parsing accuracy across multiple device models and software versions.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Imagine the human eye as a high-tech camera that takes incredibly detailed, cross-sectional pictures of the retina, the light-sensitive layer at the back of the eye. Doctors use these pictures, called Optical Coherence Tomography (OCT) scans, to spot diseases like glaucoma or macular degeneration before they cause blindness. To make sense of these pictures, the camera needs to measure specific things, like how thick the retina is or how wide the optic nerve looks. However, there's a massive problem: the big companies that make these cameras often lock their data inside secret, private formats. It's like buying a toy that only works with its own specific battery, or a video game that saves your progress in a code only the game company can read. This makes it hard for doctors to share data between different machines or for researchers to study thousands of scans at once without buying expensive, proprietary software.
To fix this, scientists use a universal language called DICOM, which is like a standard USB drive that any computer can read. Even better, they use a structured reporting system (TID 1500) that turns measurements into a clear, coded list of facts, similar to a digital receipt that lists exactly what was bought and how much it cost, rather than just a blurry photo of the receipt. The goal is to take the secret, locked-up data from different camera brands and translate it into this universal, open language so that any hospital or researcher can use it. This is crucial because without this translation, valuable medical data sits trapped in "silos," unable to help patients or advance science.
The Great Translator: Unlocking the Eye's Secret Language
Meet Transducin. Think of it as a super-smart, open-source translator that was built to break into two very different, very locked-up vaults of eye-scanning data. One vault belongs to Optopol, a company that stores its scans in a mysterious file format called .OPT that no one else knew how to read. The other vault belongs to Zeiss Cirrus, a different camera brand that does speak the universal language (DICOM) but hides its most important measurements (like retinal thickness) inside secret, private tags that standard software ignores.
The authors of this paper, led by Dr. Jesús Noel Jaurrieta Hinojos, didn't just guess how these vaults worked; they reverse-engineered them. It's like taking apart a complex clock to see exactly how the gears turn, then building a new machine that can read the clock's time and write it down in a language everyone understands.
Cracking the Optopol Code
For the Optopol .OPT files, the team discovered that the data is stored in "chunks," like a series of labeled boxes inside a suitcase. They found that one specific box, labeled OCTPARAMS, holds a secret key: a number that tells the computer which eye the scan belongs to. Here's the clever part: the number is positive for the left eye and negative for the right eye. It's not a label written by a human; it's a mathematical sign based on the geometry of the eye itself. The authors tested this on 18 different files and found it was 100% accurate. This means the software can automatically know if a scan is for the left or right eye without anyone having to type it in, even if the file name is messy or missing.
They also figured out how to tell what kind of scan it is just by looking at which "boxes" are present. If a box called ANGPRV is there, it's an angiography scan (looking at blood flow). If a box called DMARKERS is there, it's an optic nerve scan. This logic worked perfectly across 475 files from two different clinics and four different software versions spanning eight years.
Unlocking the Zeiss Zeal
For the Zeiss Cirrus machines, the challenge was different. The images were already in the universal format, but the numbers were hidden. The team found that the measurements were tucked away in private tags (like a hidden compartment in a suitcase) using a secret code called 99CZM. They decoded this secret language and pulled out the numbers for retinal thickness and nerve fiber layers. They then packaged these numbers into a "Structured Report" (TID 1500) that uses a standard medical code (SNOMED-CT) for every single measurement. Now, instead of just a picture, the computer has a clear, coded list of facts: "Retinal thickness is 283.9 micrometers," "Optic nerve cup-to-disc ratio is 0.56."
The Results: A Universal Key
The team tested their new translator, Transducin, on a huge pile of real-world data.
- Optopol: They processed 387 files from one clinic and 88 from another. The software successfully read every single file (100% success rate). It generated standard medical reports for the scans that had measurable data. For the scans that didn't have the necessary data stored (like some wide-field photos), it correctly skipped them, showing that it knows the difference between a missing file and a missing measurement.
- Zeiss: They tested 41 studies. The software successfully generated reports for all 41 of them. The measurements it found (like retinal thickness ranging from 203 to 630 micrometers) matched what doctors saw on the original screens.
Why This Matters
The paper shows that Transducin is the first tool to do this for both Optopol and Zeiss devices. It proves that you don't need the original manufacturer's expensive software to get the data out. By turning these secret formats into open, standard reports, the tool allows hospitals to store all their eye scans in one place, regardless of which machine took the picture. It also means researchers can finally combine data from different devices to study diseases better.
The authors are very sure about their findings because they tested the software on real patient data (which was anonymized to protect privacy) and compared the results directly against the original machine screens. They found that the measurements matched almost perfectly, with differences of less than 5 micrometers (which is smaller than a human hair). They also confirmed that the software works on files from eight years ago and files from the newest software, proving that the "secret code" hasn't changed much over time.
In short, Transducin is a digital key that unlocks the door to eye data that was previously stuck behind closed doors, turning private, unreadable files into open, standard information that can help doctors treat patients and scientists discover new cures. The code is free and open for anyone to use, ensuring that this translation tool stays available to the whole medical community.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.