Multi-stack registration of single-channel confocal images

[wmacturk]

Hi all,

I recently took some confocal Z-stacks of the exact same slide and XY
position in two separate channels (i.e. I found my ROI in the slide, took a
Z-stack with the 488 laser, then switched to the 405 laser and took the same
Z stack). The fluorophores were DAPI and eGFP, and I was concerned there
would be some fluorophore coupling -- the DAPI emission overlaps with the
eGFP excitation -- so I did the eGFP first, saved that stack, then ran the
DAPI and saved it to a different file. There's some XY drift in the sample
for whatever reason, with the DAPI stack obviously pushed to the right by
quite a bit, and even within the stacks, there is some drift as you traverse
the Z axis. Using stackreg it's no big deal to align the stacks to
themselves, but when using multistackreg, the result is still not aligned. I
believe the problem is that the eGFP stained bright everywhere except the
chromatin, and the DAPI stained in the opposite way. One of the cells seemed
to be undergoing mitosis, so the green channel shows a big green blob with a
kind of hole in the middle, while DAPI shows just a blue blob in that hole.

When running stackreg on the blue channel, most slices line up, but in some
focal planes, the plugin thinks one of the chromatin bundles corresponds to
a different bundle from a previous slice, so those slices get misaligned (by
quite a bit). I can manually go in and shift slices to roughly the right XY
position with Translate, but that's tedious and won't work when I start
using bigger stacks.

I'm using FIJI, and would like to automate this process fully. I'm sorry if
the question is vague, but can anybody help out? Happy to provide examples
and further info as needed.

V/R



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[CARL Philippe (LBP)]

Dear William,
Please look at the following discussion:
http://imagej.1557.x6.nabble.com/registering-a-stack-advice-needed-td5020546.html
My best regards,
Philippe

Philippe CARL
Laboratoire de Bioimagerie et Pathologies
UMR 7021 CNRS - Université de Strasbourg
Faculté de Pharmacie
74 route du Rhin
67401 ILLKIRCH
Tel : [phone]

----- Mail original -----
De: "wmacturk" <[hidden email]>
À: "imagej" <[hidden email]>
Envoyé: Mardi 8 Octobre 2019 19:29:35
Objet: Multi-stack registration of single-channel confocal images

Hi all,

I recently took some confocal Z-stacks of the exact same slide and XY
position in two separate channels (i.e. I found my ROI in the slide, took a
Z-stack with the 488 laser, then switched to the 405 laser and took the same
Z stack). The fluorophores were DAPI and eGFP, and I was concerned there
would be some fluorophore coupling -- the DAPI emission overlaps with the
eGFP excitation -- so I did the eGFP first, saved that stack, then ran the
DAPI and saved it to a different file. There's some XY drift in the sample
for whatever reason, with the DAPI stack obviously pushed to the right by
quite a bit, and even within the stacks, there is some drift as you traverse
the Z axis. Using stackreg it's no big deal to align the stacks to
themselves, but when using multistackreg, the result is still not aligned. I
believe the problem is that the eGFP stained bright everywhere except the
chromatin, and the DAPI stained in the opposite way. One of the cells seemed
to be undergoing mitosis, so the green channel shows a big green blob with a
kind of hole in the middle, while DAPI shows just a blue blob in that hole.

When running stackreg on the blue channel, most slices line up, but in some
focal planes, the plugin thinks one of the chromatin bundles corresponds to
a different bundle from a previous slice, so those slices get misaligned (by
quite a bit). I can manually go in and shift slices to roughly the right XY
position with Translate, but that's tedious and won't work when I start
using bigger stacks.

I'm using FIJI, and would like to automate this process fully. I'm sorry if
the question is vague, but can anybody help out? Happy to provide examples
and further info as needed.

V/R



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[wmacturk]

CARL Philippe (LBP) wrote
> Dear William,
> Please look at the following discussion:
> http://imagej.1557.x6.nabble.com/registering-a-stack-advice-needed-td5020546.html

Thanks for the reference Philippe. Unless I'm missing something, though, the
discussion doesn't seem to address my issue. I believe the main problem is
that, as these are confocal images, and the nucleolus is not a perfect
sphere (rather oblong, and imaged at an angle), the top slice shows a
portion that is noticeably up-and-right from the bottom slice, with no other
landmarks for registration. I'll attach two slices to show what I mean.

<http://imagej.1557.x6.nabble.com/file/t382446/DAPI_top.jpg>
<http://imagej.1557.x6.nabble.com/file/t382446/DAPI_bottom.jpg>



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[CARL Philippe (LBP)]

Dear William,
The advice I wanted to give you is that in the case the stagreg plugin is almost doing the job you are looking for, than you should give an alterative try to both the swift and template matching plugins.
Nevertheless, if I would be walking in your shoes (and I apologize very sincerely in advance if you think that what is going to follow is rude, but given that I'm french it is quite in my nature, isnt it? Thus, I hope you may forgive me at least a bit!) I would just take your data and simply trash them!!!
Once done, I would sit back on the microscope and acquire new pictures but not following the workflow you described but an alternative one...
Indeed, if you want to do something "clean", you should rather acquire the different channels as childs of your Z-stack to actually be sure not to get the drift effects you are reporting.
Also I suppose that you are not using an air objective, but rather an imersion (oil, silicon, glycerol,...) obective.
And given that these imersion media have a quite high viscosity, it is better to make your Z-stack acquisition from top down rather than from bottom up (and if you make multiposition acquisitions, adding waiting time for the imersion media to stabilize may be needed).
Respecting these rules I make the bet that your new acquired picture won't anymore need the realignments you reported.
Good luck and make our planet great again!
My best regards,
Philippe

PS: As I'm writing this answer I have a 24 hours (acquisition every 10 min) > 17 x-y positions > 5 heights Z-stack > 3 channels acquisition of a wound healing running...


Philippe CARL
Laboratoire de Bioimagerie et Pathologies
UMR 7021 CNRS - Université de Strasbourg
Faculté de Pharmacie
74 route du Rhin
67401 ILLKIRCH
Tel : [phone]

----- Mail original -----
De: "wmacturk" <[hidden email]>
À: "imagej" <[hidden email]>
Envoyé: Mercredi 9 Octobre 2019 02:37:25
Objet: Re: Multi-stack registration of single-channel confocal images

CARL Philippe (LBP) wrote
> Dear William,
> Please look at the following discussion:
> http://imagej.1557.x6.nabble.com/registering-a-stack-advice-needed-td5020546.html

Thanks for the reference Philippe. Unless I'm missing something, though, the
discussion doesn't seem to address my issue. I believe the main problem is
that, as these are confocal images, and the nucleolus is not a perfect
sphere (rather oblong, and imaged at an angle), the top slice shows a
portion that is noticeably up-and-right from the bottom slice, with no other
landmarks for registration. I'll attach two slices to show what I mean.

<http://imagej.1557.x6.nabble.com/file/t382446/DAPI_top.jpg>
<http://imagej.1557.x6.nabble.com/file/t382446/DAPI_bottom.jpg>



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html