Imaging Data Commons release v15 brings you Visible Human Dataset among several new collections, along with the improvements to the portal to simplify download process and new tutorials. See the details in the release announcement in the IDC forum: IDC July 2023 release - Announcements - Imaging Data Commons.
10 Likes
As a former member of the Visible Human Project, I’m glad to see that the data is being opened up and made more widely available.
5 Likes
I find it funny that IDC publishes it now and in such a horrible quality…
We have been doing it for like 3 years already and we have methods that don’t even require lots of ram:
Here is a short video of how snappy it feels and how clean the data is:
We used A.I. to clean up the differences and fix some errors that were inevitably there during the creation of the dataset…
everything you see is live and not rendered… it’s real time
I am thinking about asking our team leader to publish this data and to also publish methods hwo to use it without lots of ram. We are a private company and we never published anything academically so we would need some advice on how to do it appropriately and where
We in IDC strongly believe there is always value in making the original data accessible. This way we can have more users innovate and come up with better and better ways to process and improve it.
What we showed in the video is also real time.
If you ever do decide to publish it, we in IDC would be happy to review your contribution and consider hosting it alongside the original data. Sharing the original VHD dataset was always meant to be the beginning, not the end - we welcome contributions that enrich or improve the original VHD dataset.
But you must publish your dataset under a compatible license for it to be in IDC, and most of our data is available under CC-BY or comparable licenses.
The (in your words) “horrible quality” original VHD dataset is available to everyone under NLM license that has very few restrictions, and does not forbid commercial use. “C” in IDC stands for Commons.
As far as I know the VHP Female data digitized from the 70mm film photos is the highest quality available. It is very interactive to view if you have a big enough machine:
It would be great to hear if people know of higher quality data. Or non-human data like this of high quality could also be interesting to look at.
I saw lots of errors in the male dataset: distortions, misregistrations, missing frames, blank frame, color differences between sections, etc. in the optical images; and overall low quality (by today’s standards), some discontinuities, etc. in the radiology images. I remember that I was quite disappointed, too, but on second thought it was understandable that mistakes were made, as it was not a routine operation (such project had not been done before).
So, I would say that the quality of the raw dataset was actually reasonable. The “horrible” thing was that the data was encumbered by a restrictive license for many years. The license prevented people from reprocessing and fixing up the raw data and redistribute the improved results.
The current permissive license and better accessibility via IDC will hopefully motivate more groups to use the data, improve it, and contribute back the improvements to the community.
I apologize to everyone for writing the word “horrible quality”. It was definitely a wrong word to use. This what they did is anything but horrible… I was in my bed, sleepy and I was online, didn’t really think my words through.
The CEO of our company attempted to purhcase datasets from the Korean and Chinese projects, they were made later and form what we understood, they have higher resolution. They rejected the proposal and clearly what ICD does is fixing that problem exactly. You guys made it available to everyone and that’s amazing…
1 Like
Blockquote
Yes I agree 100%
We wanted to establish a small lab where we could have this process in a streamlined and a systematic way and then create sets of images of animal and plants anatomy for download.
The issue: 1. regulations and 2.costs
Incredible number of regulations and restrictions were the primary cause that prevented our company from establishing such a lab.
Technically we were sure that animals could be processed the same way but plants? I’d like to see some attempts…
Here are some interesting sources, I wish they would make the Korean Visible Human project available the way IDC made this possible: Visible Korean human: improved serially sectioned images of the entire body - PubMed