Hello everyone.
Question.
I currently use OHIF as a viewer, and in Orthanc I compress the images using “IngestTranscoding”: “1.2.840.10008.1.2.4.90”. While the files aren’t exactly large, some CR images are 4 to 6 MB and some MG images are 6-7 MB.
To preserve the original DICOM data and avoid any loss, this works perfectly.
But I have the problem that sometimes some locations have a poor internet connection, and this slows down the loading time, especially if there are many images. It’s not a huge delay, but it can be noticeable.
I would like to have a faster version to view the same images, and that’s where I’m having trouble finding a solution.
Some colleagues, I don’t know how they do it, have a very small preview image (an image that originally weighs 18 MB in .90 format) but they have a version that weighs only 700 KB. I don’t know how they do it, but they use Oviyam and I assume some other method (I tried the latter, and for example, the same image compressed to .90 in OHIF, viewed with OHIF is 9 MB, and viewed with Oviyam it was 3.7 MB, but with Oviyam the window level didn’t work; it only recognizes it if the images are stored in Orthanc with the “1.2.840.10008.1.2.1” format).
In my case, I thought about resizing the images, which I managed to do, but I don’t know if generating a new study with the reduced version would complicate future queries to Orthanc, because imagine, if I have 50 studies (CR and MG), I would have a total of 100 studies per day, and this would double every day.
I also thought about adding a new series with the reduced images, but it’s complicated with the description (because each CR image has a separate series, so one description per series).
That’s why I’m asking, how can I do this? Is there any viewer I can use with Orthanc that can display a resized or scaled image, or one that’s much lighter in file size than what I currently have? Ideally, I’d like to configure it to send me the image in the format I need. I’m really stuck on this, and it’s causing me problems. Any help would be greatly appreciated.
Using "IngestTranscoding": "1.2.840.10008.1.2.4.90" (JPEG2000 lossless) preserves the original DICOM data but does not significantly reduce file size. For much smaller images, you would need lossy compression (e.g., "1.2.840.10008.1.2.4.91" – JPEG2000 lossy or "1.2.840.10008.1.2.4.50" – JPEG baseline), but this sacrifices diagnostic fidelity.
Orthanc supports on-the-fly transcoding on retrieval (FAQ on transcoding). This means you can always store the lossless originals, but configure Orthanc to serve lighter images when requested by the client (e.g., OHIF). This avoids duplicating studies and keeps storage under control.
Another option is to generate derived series with resized images, but this makes study organization more complex (especially for CR, where each image can already be a separate series).
Some viewers (like Oviyam) appear faster because they often display JPEG thumbnails instead of full DICOM images, which explains the smaller size but also the lack of features like proper window/level. For quick previews on slow networks, you could use a lightweight viewer such as the Orthanc WebViewer plugin, while still keeping OHIF for diagnostic use.
In short: the most practical solution is to store originals in lossless format and enable on-the-fly lossy transcoding for faster visualization when needed. This way, you balance diagnostic fidelity with performance, without duplicating studies.
Answer always based on the Official documentation in Orthanc Book. But of course, I could be wrong.
Dear Lucas, thank you for your recommendations and explanation.
Here are my comments:
With the Orthanc web viewer, I noticed that the images in low resolution are requested using the URL “http://localhost:8042/web-viewer/instances/jpeg80-{id_instance}_0”. This way, a very small image is downloaded, which is what I need. The problem is that it doesn’t display the entire study, nor a preview, and besides, I have to click the “L” button on the toolbar to get it to download in that resolution.
With OHIF, I couldn’t request the images in a lightweight format as in point 1; I could only work with the original images in .90 format. This would have been perfect if I could request them as the web viewer does, because it would work much faster.
I tried to make on-the-fly requests, but without success. I couldn’t get it to work, and I don’t understand why.
I tried to create something with Cornerstone and React, but I can’t get it to work or load the images.
Now I’m looking for a simple viewer that I can modify, but that makes requests like the Orthanc web viewer. I’m trying to read more to understand it better, but it takes a lot of time, and sometimes it confuses me.
Any further recommendations would be appreciated.
Best regards
In short: the most practical solution is to store originals in lossless format and enable on-the-fly lossy transcoding for faster visualization when needed. This way, you balance diagnostic fidelity with performance, without duplicating studies.