general
5 Subscribers 
general > Getting original vs. transformed cell coordinats
Showing 1-3 of 3 posts
Jul 5, 2020 02:07 AM | Boaz Mohar - HHMI
Getting original vs. transformed cell coordinats
Hi,
First, thank you for these amazing tools.
I have used both QhickNII and VisuAlign to get a great registration of my imaging stack to the mouse Allen CCF.
I then used Nutil to feed in a segmentation image (from Ilastik) and got the right results in terms of looking with the online MeshView
Attached is an example (slide1.png).
However I have a problem, I have other information relating to each one of these cells, and I want to know which one of the cells in the output report is which from the input.
Is there a way to know what is the order in which cells were detected?
Can I feed a label image which each cell having a different number to keep identity?
As I have already thresholded and trained a second Ilastik object based network I don't want to lose that information.
Any advice would be appreciated.
Thanks again,
Boaz
First, thank you for these amazing tools.
I have used both QhickNII and VisuAlign to get a great registration of my imaging stack to the mouse Allen CCF.
I then used Nutil to feed in a segmentation image (from Ilastik) and got the right results in terms of looking with the online MeshView
Attached is an example (slide1.png).
However I have a problem, I have other information relating to each one of these cells, and I want to know which one of the cells in the output report is which from the input.
Is there a way to know what is the order in which cells were detected?
Can I feed a label image which each cell having a different number to keep identity?
As I have already thresholded and trained a second Ilastik object based network I don't want to lose that information.
Any advice would be appreciated.
Thanks again,
Boaz
Aug 21, 2020 07:08 AM | Sharon Christine Yates - University of Oslo
RE: Getting original vs. transformed cell coordinats
Hi Boaz,
Sorry for our late response, we didn't pick up on your request until now.
I have transferred your issue to the github page for Nutil. You can follow the request here:
https://github.com/leuat/nutil/issues/14...
I will discuss this with our programmer and get back to you ASAP.
Best wishes,
Sharon
Sharon Yates
researcher,
Nesys Laboratory,
University of Oslo
Sorry for our late response, we didn't pick up on your request until now.
I have transferred your issue to the github page for Nutil. You can follow the request here:
https://github.com/leuat/nutil/issues/14...
I will discuss this with our programmer and get back to you ASAP.
Best wishes,
Sharon
Sharon Yates
researcher,
Nesys Laboratory,
University of Oslo
Aug 21, 2020 11:08 AM | Sharon Christine Yates - University of Oslo
RE: Getting original vs. transformed cell coordinats
Hi there, and thanks for this excellent question/remark!
The easy way: If you only need to "keep track" of the various areas, it would be quite an easy task for me to a) create a label for each identified area b) print out the ID of the area on the visual output images (where you've already got a number for the region ID etc) and c) list this ID in the reports. Actually, this is an excellent idea - and the whole thing could be enabled by having a single "yes/no" for ID Tracking in the options list.
The hard way: if *you* want to track the IDs (that is, by coming up with a name and assigning that to a specific area), I'm really uncertain how this would be implemented - and whether it is feasible to do at all. I therefore suggest we go for the easy way first. How does this sound?
btw: all requests/bugs can be tracked through the github issues: https://github.com/leuat/nutil/issues/14... so please use github for future communication.
Nicolaas
The easy way: If you only need to "keep track" of the various areas, it would be quite an easy task for me to a) create a label for each identified area b) print out the ID of the area on the visual output images (where you've already got a number for the region ID etc) and c) list this ID in the reports. Actually, this is an excellent idea - and the whole thing could be enabled by having a single "yes/no" for ID Tracking in the options list.
The hard way: if *you* want to track the IDs (that is, by coming up with a name and assigning that to a specific area), I'm really uncertain how this would be implemented - and whether it is feasible to do at all. I therefore suggest we go for the easy way first. How does this sound?
btw: all requests/bugs can be tracked through the github issues: https://github.com/leuat/nutil/issues/14... so please use github for future communication.
Nicolaas