Multi-channel Multi-view Deconvolution: Fusing Second Harmonics Channels

Hi everyone,

Before I start, I want to thank this community for the help that it has already provided me. I have lurked for some time here but an issue in my current project motivated me to start this thread.

I am using a two photon fluorescence microscopy to make multiview reconstructions of EGFP labeled zebrafish. I have images of two channels: the emissoins from the EGFP and the one from secon harmonics generations. I can easily fuse the EGFP channel because I added fluorescent beads to the agarose gel in which the zebrafish are embedded. Unfortunately, the second harmonics channel -which does not detect the beads fluorescence- stays unmutated by the Multiview Deconvolution Plugin.

I am aware that the ImageJ page ( outlines a method to translate the deconvolution coordinates from one channel to another channel. However, my knowledge in Java is too basic to follow this method.

Instead, I have tried fusing both channels toguether (following the method here outlined: to trick Fiji into fusing them toguether. However, it seems to be able to recognize that there is more than one channel and excludes the second harmonics form the multiview deconvolution.

I was wondering if someone knew of a way to ‘fuse’ two channels in a way that tricks the Multiview Deconvolution Plugin.

Thanks in advance!


1 Like

@StephanPreibisch - would you be able to assist with this particular issue?


Hi, duplicating transformations from one channel to the other is rather simple:

  1. you only need the saved XML that contains the correct transformations for one of the channels.
  2. Open Plugins > MultiView Reconstruction > Batch Processing > Tools > Duplicate Transformations
  3. Choose “One Channel to other Channels”
  4. Choose the XML
  5. Press OK
  6. Select the source channel (the one that has the correct transformations), leave the rest as-is (replace all transformations, All channels as target)
  7. Press OK
  8. Open it in the MultiView Reconstruction or BigStitcher and it will be done
1 Like

Thanks so much. I had found an ulterior way but it was a bit lengthier. I was just fusing the fluoresent channel containing the beads and then switching the names of the fluorescent channgels with those of the second harmonic channels (that had not beads). Therefore, when deconvoling the dataset, the Multiview deconvolution application would use the second harmoncs channel instead of the fluorescent one. I found it worked fine in all cases which I have tried.