Guide

How to capture endoscopy images on a computer

A room that documents well is not the one with the most expensive tower. It is the one where the signal path is right, the images are saved the moment they are taken, and somebody decided in advance which landmarks get photographed. This guide covers all three, and it applies whatever software you end up using.

Clinical review: Dr. Omar Quiroz, Gastroenterology and Endoscopy, Endoscopia del Mayab, Mérida.

1. The signal path

An endoscopy processor produces a video signal and sends it to a monitor. To get that same image onto a computer you take a second output from the processor and feed it into a capture card, which the computer sees as a camera. That is the whole idea; everything else is choosing connectors.

• Prefer a digital output. HDMI, DVI-D and SDI all carry the image without loss. An analog output (S-Video, composite, RGB) gives you standard definition no matter what you connect it to.
• Use a UVC capture card. A card that follows the USB Video Class standard needs no drivers on macOS, which means one less thing to break after an update.
• Keep the cable run short. HDMI becomes unreliable past roughly 10 metres. SDI does not have that limit, which is why operating rooms use it.
• Don't share a hub. Plug the capture card straight into the computer. A hub shared with disks and a keyboard is a common cause of dropped frames.

The compatibility page lists the outputs found on the processors we are asked about most, connector by connector.

2. Resolution, frame rate and what actually limits quality

Capture at the resolution the processor sends, usually 1080p, and do not upscale. Nothing downstream can add detail that the signal never carried.

Frame rate matters more than people expect, and not for smoothness. At 60 fps the capture card is handing the computer sixty chances per second to find a frame that is in focus. Motion blur is what ruins endoscopy stills: the mucosa moves, the scope moves, and the frame that arrives at the instant you press the button is very often not the sharp one. Software that keeps the sharpest of the last few frames, instead of the last one, is the difference between a usable photo and a smear — and it costs nothing, because those frames already went past.

Save stills in a lossless or lightly compressed format. Heavy JPEG compression puts blocking artefacts exactly where you are trying to judge a subtle mucosal pattern.

3. Which images to document in each procedure

Photodocumentation is easier to defend when it follows a fixed list of anatomical landmarks rather than the operator's instinct in the moment. The European and American endoscopy societies (ESGE and ASGE) both publish photodocumentation recommendations; the lists below are the practical version of that idea and the one worth agreeing on with your team.

Upper endoscopy

• Upper and middle esophagus.
• Z line and the diaphragmatic pinch, with the distance in centimetres noted.
• Gastric body along the greater and lesser curvature.
• Incisura angularis.
• Antrum and pylorus.
• Fundus and cardia on retroflexion. This is the view most often missing, and the one where lesions hide.
• Duodenal bulb and second portion, with the major papilla.

Colonoscopy

• Cecum, documented. The appendiceal orifice and the ileocecal valve in the same frame if you can. This is the image that proves the study was complete.
• Terminal ileum, when intubated.
• Ascending, transverse, descending colon and sigmoid.
• Rectum on retroflexion.
• Every lesion: before, during and after any intervention, with something in frame to give scale.
• Bowel preparation, if it was poor. A photo settles the argument about whether a repeat study was needed.

CPRE

• Major papilla before cannulation.
• Fluoroscopic cholangiogram.
• The therapeutic step: sphincterotomy, stone extraction, stent placement.
• Control cholangiogram at the end.

4. Saving: the part that goes wrong

The failure that costs a room a whole morning is almost never the capture card. It is software that holds images in memory and writes them at the end, so a power cut, a crash or somebody closing the wrong window takes the procedure with it.

What to ask of whatever software you use:

• Each image is written to disk at the instant it is captured, not when the study is closed.
• After a crash, reopening offers to recover the procedure with its images rather than starting a blank one.
• It is difficult to leave an active study by accident.
• The study is usable with no internet connection. Connectivity should be needed to share, not to work.

This is the behaviour EndoMac Pro was built around, and you can try the failure on the home page: capture a couple of images in the simulator and trigger a power cut.

5. Sizing the archive

Stills are small. A 1080p endoscopy image is on the order of one to three megabytes, so a study with twenty captures is tens of megabytes and a busy room doing ten studies a day produces a few hundred megabytes a day — a couple of hundred gigabytes a year, comfortably inside one disk.

Video is the opposite. Recording full procedures in HD runs to gigabytes per procedure and will fill any disk you give it. Record clips of the moments that matter rather than everything, and decide deliberately how long clips are kept.

Two copies, in two places, is the minimum that deserves the name backup: the Mac in the room for work in progress, and a NAS in the clinic for finalized studies. A NAS with mirrored disks survives a disk failure; it does not survive a fire or a theft, so if the archive is the clinic's legal record, a third copy off-site is worth the trouble. Test a restore once. An untested backup is a belief, not a backup.

6. Patient data in the image

Many processors burn the patient's name and the date into the video output. Anything burned in is captured with the image and travels with it — into the report, into a shared link, into whatever the patient forwards. Decide on purpose what the processor should overlay, and keep identifying data in the record rather than in the pixels, where it can be controlled.