Using KymographBuilder Plug-in does not show output image if using PyImageJ in Python

[JeBiUKD]

Hi,

I am currently working on a skript in phython where I integrated a ImageJ macro using PyImageJ. I have been able to work with previous plug-ins like the Multi Kymograph plug-in (https://github.com/fiji/Multi_Kymograph). However, for my analysis I get a better output if using the KymographBuilder plug-in (https://github.com/fiji/KymographBuilder).

The problem which I am facing is that it successfully runs the KymographBuilder plug-in (see below) however does not show an output image/window at the end, which I need in order to save the generated kymograph.

Also when I am running the macro in Fiji/ImageJ it works without any problem and shows me the desired output image. So it hast something to do with the PyImageJ integration.

I am not sure if its relevant but I am working with Jupyter notebook.

[java.lang.Enum.toString] [WARNING] Ignoring line width >1. [java.lang.Enum.toString] [INFO] Running KymographBuilder version 3.0.0 [java.lang.Enum.toString] [INFO] Dataset : stack_all.tif Width = 2532 Height = 2132 Depth = 1 Timepoints = 118 Number of channels = 1

[java.lang.Enum.toString] [INFO] 1 lines with a width of 100 will be used for the kymograph 0. [java.lang.Enum.toString] [INFO] Creating kymograph for the channel. [java.lang.Enum.toString] [INFO] Kymograph "stack_all.tif (Projected Kymograph)" has been correctly generated.

I hope you can help me find a solution with this specific issue.

Thank you in advance!

BTW I submitted the issue here as well, as I am not sure who can adress it (https://github.com/fiji/KymographBuilder).

[elevans]

Hi @JeBiUKD,

It sounds like you are running in headless mode when instead you want interactive. How are you initializing PyImageJ? By default you get headless mode when you initialize PyImageJ like this:

import imagej

ij = imagej.init()

To get interactive mode initialize PyImageJ like this:

import imagej

ij = imagej.init(mode='interactive')

Once in interactive mode you can call the ImageJ/Fiji GUI and view images. This only works with jupyter notebook locally.

For more info on initializing PyImageJ, check out the docs here.

[JeBiUKD]

Hi,

thank you very much for the response. I used this to initilize PyImageJ:

ij = imagej.init('sc.fiji:fiji',headless=False)

I also tried your method (ij = imagej.init(mode='interactive')) and it gives me actually a different error message:

Unrecognized command: "KymographBuilder" in line 13
run("KymographBuilder",input=stack_0_100.tif"<)>;

It looks like it cannot find the plug-in. The same happens when calling the previous plug-in which worked before in ij = imagej.init('sc.fiji:fiji',headless=False). So it seems I cannot call plug-in at al by run("xxxx");

Any ideas?

Best, Jessica

[elevans]

Hi @JeBiUKD,

Sorry for the delayed response, I was out of town teaching.

What version of PyImageJ are you using? The headless flag has been deprecated and replaced with mode. So if you want the latest a interactive Fiji you'd initialize ImageJ with ij = imagej.init('sc.fiji:fiji', mode='interactive'). When you initialized ImageJ with ij = imagej.init(mode='interactive') you only get ImageJ2 and not Fiji, which is why you get an error indicating the KymographBuilder plugin is missing.

Here's a minimal example that uses the KymographBuilder plugin on some sample images in Fiji.

import imagej
import code

# initialize Fiji in interactive mode
ij = imagej.init('sc.fiji:fiji', mode='interactive')

# display ImageJ GUI
ij.ui().showUI()

# open sample data and run KymographBuilder
ij.IJ.run("Confocal Series")
ij.IJ.makeLine(136, 125, 267, 236)
ij.IJ.run("KymographBuilder", "input=confocal-series.tif")

# return REPL
code.interact(local=locals())

Can you also try just adding ij.ui().showUI() to your original workflow? When I was playing around with this I noticed that the output from the plugin didn't display until I showed the ImageJ UI. There might be a bug in the plugin that ties it's output to some other display elements associated with the main UI.

[JeBiUKD]

Hi @elevans,

Adding ij.ui().showUI() did the job.

Thanks so much for your time and the easy solution.

Best, Jessica