(part 1) Step 4: averaging BB per cell
For this step, make sure that the AveragingND >= 0.1.0 is installed on your FIJI.
There is no separate macro for this step. During batch processing batch_many_folders_.ijm runs this plugin
and creates new s3_rotated_XY[scale]_Z[scale]_avg folder and stores the output there.
So if you run it manually, you need to create an output folder.
Open FIJI and run Plugins->AveragingND 0.1.0 -> Iterative Averaging command.
A new window will appear, where you need to specify how input data will be loaded.
We will load tiff files from the folder. The choice of low/high memory
depends on the available RAM, if it can fit all files.
After clicking OK a new folder-choosing dialog will appear,
you need to select s3_rotated_XY[scale]_Z[scale] folder from the previous steps,
containing straightened and rotated tif volumes.
After that iterative averaging parameters need to be specified:
where:
- first parameter specifies the alignment channel;
- second option defines how volumes are aligned for the first, initial template/iteration;
- total number of iterations to run;
- Template calculation: simple average or an option where voxels with zero intensity are not included in the average;
- whether to use zero-masked normalized cross-correlation (i.e. exclude voxels with zero intensity);
- registration constraints (maximum shift of volumes in each axis);
If the "Intermediate averages?" option is selected, the average volume at each iteration will be saved in the intermediate subfolder.
If the "Register inputs?" option is selected, the plugin will output all volumes with the final registration.
In the last dialog, you need to choose an output folder.
The main output is final_avg_[iter number].tif file containing
the final averaged volume with the suffix of the iteration having the best (maximum) cross-correlation.
The last step is (part 1) Step 5: average BB shape quantification
Using "total protein" staining as a reference channel,
at the initial step (iteration 0) the plugin creates a reference "template" image
as an average of all input images. Images are positioned so their centers are aligned
and padded with zeros.
After that, it will iteratively register input images to this average.
Once it is done, a new average is built with updated positions (iteration 1)
and the cycle continues until user-specified number of iterations is reached.
The movie below shows evolutions of the updated average.
