Is it possible to embed a custom DICOM PACS viewer in prodigy web interface to annotate on medical data which can contain may slices and enable annotation per slice.
Do you have an example of how a viewer like this would look like, and what exactly you’d be annotating here?
In general, Prodigy allows you to stream in images and other data from any source – if you can load your images in Python, you can annotate them with Prodigy. You can also use a custom HTML interface if you need more flexibility. The usage workflow on computer vision also has an example of how to write a simple, custom recipe that streams in images and lets you annotate whether a label applies to it.
This is how typical PACS Viewer looks like above . A radiologist would look at slices of images and answer series of questions ( as part of annotation).
Thanks a lot – this is really interesting!
I think the answer I just posted on this thread already goes in a similar direction – essentially, all you need to do is export your images and add your questions to them programmatically.
Keep in mind that Prodigy’s annotation interfaces are intended to let the annotator move through examples very quickly and focus on one decision at a time. So you can use this to your advantage by framing each decision as an individual task that can be answered by one or two clicks or key presses. You might even be able to reframe some of the questions to make them even quicker to annotate – for example, by phrasing them as binary decisions. Ultimately, a lot of this will come down to experimentation – and being a little creative 
The use case you describe – i.e. having domain experts like radiologists doing the annotation work – is definitely something we’ve had in mind when building the tool. The main focus here is getting most out of the annotators time, helping them focus on what really matters and let the computer handle the repetitive and tedious aspects of the annotation work wherever possible.
I am planning to build a whole annotation product using spacy, and prodigy for healthcare. Will reach out to your team in next week to understand how to have UMLS, Radlex, Snomed concepts integrated with Prodgy. Looking forward to working with your team.