Using Fiji's multipoint tool to create 2 different labels

fiji
imagej
fish
multi-point

#1

Hi,

I have some images from a FISH experiment in which cells contain 2 different types of signals (for my locus of interest and a reference) in different colours/channels. I’ve attached a couple of images as examples.

I’m scoring this manually, which is relatively straightforward. I think the point tool in FIJI (2.0.0. on Mac) could be useful to quickly count and save a copy of the analysis. However, as far as I’ve seen I can only create a series of points. Does anyone have a workaround or suggestion to be able to create 2 point groups (with different point types and counters) on the same image? I can obviously do it in separate copies of the image but it’d be useful to have both overlayed.

I don’t know if this something the multi-point option could be used for - I haven’t been able to make much sense of it and the tool seems to have changed compared to what’s described in the user manual I’ve found online (from FIJI 1.46.).

Any tips would be much appreciated.

Thanks!

FISH2|nullxnull FISH1|nullxnull


#2

Good day,

what exactly do you like to achieve?

If you get the labels set correctly what do they indicate?

It is no problem to automatically record the positions of all red and green markers separately but what for?
Do you need their relative positions or their positions with respect to the blue “clouds”?

In any case, using the ROI-manager you can set labels having different colors in the same image.

Regards

Herbie

PS:
Here is an example of automatic labelling:


Green markers are labelled by yellow dots and red markers by cyan crosses.
(The weaker markers could be labelled as well.)


#3

The options of the multi-point selection tool allow to specify different counter types. Double-click on the multi-point tool to open the options dialog:

image


#4

Hi Jan,

Thanks! First time using the tool and I hadn’t quite figured it out. This is what I had in mind. I see one can only alter the colour (not type) of the label for each counter type, but that’s more than enough.

Regards,

Carlos


#5

The colors are actually hard-coded, i.e. each counter has its color and cannot be changed. The Type and Color options on the top are only valid for the “simple” multi-point tool, but are ignored when used in multi-counter mode…


#6

Hi Herbie,

I’m really just looking at quantifying both types of signals. Relative positions are irrelevant and number of each within each nucleus (the blue clouds you mentioned) is not that important, since averages across all the cells in the image give an average count more representative of possible heterogeneity.

Thanks for the tip on using ROI-manager to change label properties. I think counting of the two different types of labels is more straightforward using the multi-point tool, but this is also useful to know as an alternative.

Regards,

Carlos


#7

You’re right, so they are! Although it doesn’t really matter when the only aim is to differentiate labels.

Thanks again!


#8

Carlos,

when you write

I’m really just looking at quantifying both types of signals.

you mean just the number of green and red marker spots in the image?

Could you please tell me why you need labels in the image if you are just interested in the numbers?

Could you please comment on the result I’ve posted in my previous post.
(The number of marker spots that were labelled is known I didn’t post it yet.)

Regards

Herbie


#9

Yes, that’s what I meant. I was simply trying out the labels as a tool to be able to count quickly and also save an image in case I need to go back to look at how a specific case was scored. This is not strictly necessary, I just wanted to figure out how the point/multi-point tool could be used.


#10

Carlos,

in case I’d provide a macro that does automatic labelling and counting, would you still prefer doing it by hand and why?

The image I’ve posted before was the result of automatic labelling.
(I simply didn’t post the counts for each label.)

Regards

Herbie


#11

Herbie,

Yes, I was always aware that this could be relatively easy to automate, given how distinct most of the signals are. However, for my application of interest manual counting tends to be the norm, since there’s often signal variation (in size, split signals, etc) and other considerations not present in the examples I provided that could be harder to incorporate into a macro. I may consider doing manual scoring but also running a macro to see how both methods compare.

Regards,

Carlos


#12

I see Carlos,

in order to understand what is so special about countin by hand, would you please do the comparison (hand versus automatic) for the result I’ve provided. Otherwise it is impossible to understand what needs to be improved or to decide that automatisation isn’t worthwhile …

Regards

Herbie


#13

Just an add for the manual counting Fiji provides also a tool called Cell Counter under Plugins->Analyse->Cell Counter

where you can do something similar to the one suggested by @imagejan

Like this:

52

It’s very similar to the multipoint tools above mentioned:
you lose some options of the markers (like size and type) but it counts directly for you.

have a nice day,
HTH
emanuele martini


#14

A manual count (attached) gave me 89 green signals to 35 red signals. From a quick look at your automatic counts, those were about 67 green and 47 red. That’s arguably not a huge variation and I realise you could always adjust your settings. It should be noted that this image is an ideal example in which most signals are very clear and similar in size. I do agree that I shouldn’t rule out using a macro, which is why like I said I’ll probably use both in parallel and compare results.

Hope that makes sense.


#16

Hi Emanuel,

That’s also useful to know.
Thanks very much,

Carlos


#17

Carlos,

many thanks for the details and your judgement.

It should be noted that this image is an ideal example

This is perhaps the most important statement!
Indeed the first sample image is less clear and contains large structures that need to be suppressed in a preprocessing step to make the counts reasonable.

If you decide one time to add automatic counting and labelling then please let us know.

Best

Herbie


#18

Agree on that! I’m only getting started doing this, but will look into the different options. While I’ll probably go with the more established approach (manual) to start with, I definitely think it’d be interesting to assess how different scoring methods compare.

Regards,

Carlos