Modifying OME Data model for simultaneous image acquisition

[gbonamy]

Dear all,

I propose to modify the OME metadata model to introduce the notion of multi-channel imaging. Many new high-content imager such as the Opera, Cell Voyager, ImageXpressUltra, ... as well as some confocal microscope are able to image simultaneously several fluorescence channel.

It is always easy to associate a given image to a single detector, however, the excitation source is currently strictly associated with a single laser/dichroic. With these instruments, the model does not hold true, because a fluorofore is actually excited with several lasers (or multiple excitation sources) simultaneously.
Thus, saying GFP is exited with 488nm is not true if you also have a 405nm and 633nm laser line activated (to excite other fluorofores). I agree, that in this case 633 probably does not have a big impact on GFP excitation, but what 405 does. So I think it might be justified to allow multiple excitation sources for a single image. Similarly, since multiple dichroic are use to separate the different channels, I think there should be a way to report more than one dichroic per channel.

I hope this makes sense.

Best,

Ghislain

[ajpatterson]

Hello Ghislain,

We have had the use case of multiple light sources raised once before in relation to LED light sources.

In principle there is no problem with this. We would need to change the arity of LightSourceRef within LogicalChannel from 0-1 to 0-*. This change would not invalidate existing files so would be a Non-Breaking change.

What has not been raised before is the idea of multiple dichroics.

Again there is in principle no problem with this. We would need to change the arity of FilterSetRef within LogicalChannel from 0-1 to 0-*. This change would not invalidate existing files so would be a Non-Breaking change.

There are a couple of stumbling blocks however.

If both changes are made you cannot tell which LightSourceRef goes with which FilterSetRef. Or if they are all active at once. This means we would need to work out some method of linking them together.

Also FilterSetRef is also used in OTF (The optical transfer function) with an arity of 0-1. Would this also need to become 0-* and are any other changes required?

Thanks again for you input,

Andrew

[gbonamy]

Andrew,

Thanks for the reply and sorry for the delayed answer. The problem with using a 0-* model is that it is not necessarily true. The opera on top of offering multiple simultaneous excitations can also run multiple exposures (now only 2 separate exposures).

For instance, one could image imaging Hoechst and DsRed together (turn on 405 laser and 561). Then to avoid cross talk a second exposure with 488nm (and a different set of dichroic) would be used. All of the images collected are set into a single files, but the different "Channels" do not make use of all 3 lasers.

Thus, I would recommend a model with X-Z and V;X;Z and * (for all for instances).

For the AOTF, the Opera does not use any, but it is likely that other instrument use one AOTF per laser line. In my opinion the AOTF should be a property of the light source rather than a free standing component. If all the lasers use the same AOTF, then they would all point to the same one.

Hope this is useful,

Ghislain

[wmoore]

Hi Ghislain,

When you excite with 2 lasers, do you use 2 detectors (split by a beamsplitter) to collect 2 channels?
If so, could you link the laser and excitation filter etc to the appropriate channel?

I appreciate that this does not model exactly what is happening on the microscope, but if you are collecting 2 channels and combining with a 3rd to make a 3 channel image, then this would be the best way to model it, and it wouldn't require a change to the OME model.

Andrew is away just now.

Will.

[thomaslatfield]

I think you would be able to link the laser and excitation filter to the appropriate channel. Try using the the 2 detectors to collect 2 channel. that might work for you.

cheers, Thomas