Commit 02be655c authored by nadar's avatar nadar

added documentation from various sources

parent e66ab821
# Deconvolution with Huygens using theoretical PSF
-
Author: Gayathri Nadar, SCF MPI-CBG nadar@mpi-cbg.de </br>
Date: 13.08.2019
## Parameter wizard
- Open image and first verify the parameters. If everything is fine, all the information is taken from the metadata itself.
- You can check this via `right click > parameter editor/wizard` or `deconvolution > parameter wizard` on top menu.
- The parameters need to be set for each channel.
###Parameters
Radio buttions: White - metadata, green - verified before, black - edited, orange - default, blue - estimated
- #### Sampling density for x, y, z
- Should satisfy Nyquist sampling rate. Use https://svi.nl/NyquistCalculator to compute your sampling rate based on microscope parameters. If your actual sampling rates are far apart from calculated ones, choose a value close to the computed ones.
- Red, orange - undersampled data, not suited for deconvolution. Blue, purple - oversampling, not a problem for deconvolution but computationally expensive. White - expected range, all okay.
- #### Coverslip parameters
Visual editor is available for help.
- position (um): distance from first z-slice of stack
- light direction: upwards or downwards based on microscope
- quality of objective
- #### Optical parameters:
- numerical aperture
- refractive indices: lens immersion and medium
- #### Channel parameters:
Set for each channel
- microscope type: for multiphoton select confocal and select number of excitation photons > 1
- excitation photon count: should be > 1 for multiphoton, else 1
- emission wavelength: wavelength of light emitted by the subject
Once set for all channels, **export this as a template and save it for future use.**
## Deconvolution wizard
- #### Parameters
- This is what we did in the previous section. Make sure the parameters are verified for each channel.
- Load the saved template if you have one or enter the values.
- #### PSF
- If we already have a PSF, we can load this here.
- If nothing is selected, Huygens uses a theoretically computed PSF.
- #### Cropping (optional)
- Deconvolution is a slow process and can be made faster by cropping the volume to a smaller region.
- Once a pipeline is finalized, it can be used to run on the whole image.
- There is option `Auto crop` which automatically crops your volume.
- You can also launch the `Cropper` itself.
- The cropper displays image in three directions xy, xz and zy.
- There are also options on the right pane to select channels.
- The `Lock ratio` option locks the aspect ratio of the crop.
- The `Preview` option helps to view a MIP of the cropped region.
- Cropping can be done by dragging the orange box or by entering the values manually in the right panel. Do not crop the image too tightly, this does not help in deconvolution.
- #### Channel
- Select the channel to process here.
- For every channel an output will be created in the left panel.
- #### Histogram
- Important tool to check image quality. The peaks represent number of voxels for a particlar intensity.
- Some aspects
- High peak on right: If the height of the histogram is higher than zero on the very right of the diagram, your images are over-saturated. Continuing deconvolution is not recommended in this case. caused by intensities above the maximum digital value of the microscope. They are set to highest value possible. Clipping will have a negative effect on the results of deconvolution, especially with widefield images.
- Narrow peak at left represents background pixels
- Gap on left: black level is not zero. Deconvolution is not so suitable for this case.
- #### Background estimation
- In this step, the mean background intensity is to be estimated and entered.
- Automatic estimation
- Widefield mode: searches for 3D region with lowest values to select region with least amount of blur
- Lowest: searches for 3D region with lowest average value
- In/near object: searches for a 3D region for lowest value near high intensity object
- Area radius: size of 3D region to estimate background.
- Manual mode
- Draw a line across the images covering object and a background.
- Plot on the right shows the profile.
- Double click on a value to get background intensity. Find a suitable one which represents **maximum** background intensity. Enter the value and click accept.
- #### Deconvolution
- Enter parameters for running the deconvolution.
- **Maximum iterations**: The process runs iteratively so you need a stopping criterion which is given by the maximum number of iterations. Start with 50-80.
- **Signal to noise ratio**: The SNR is defined as the ratio between average signal intensity of an object in foreground divided by the standard deviation in the background. Using a too large SNR value might be risky when restoring noisy originals, because the noise could just being enhanced. A noise-free widefield image usually has SNR values higher than 50. A noisy confocal image can have values lower than 20. Very high value enhances the noise in the image and low value causes signal loss. Start with value such as 20 and gradually move to 40 and 60. Compare the output and finalize one option.
- **Quality threshold**: This is another stopping criterion. Once the change in quality between different iterations drops below this value, the deconvoltuion process stops. Most of the times, this happens before even the maximum number of iterations is reached. If it does not, increase the number of iterations. Leave it to default.
- **Iteration mode**: Leave it at `Optimized` mode.
- Bleaching correction: Leave it at `If possible`. This will automatically check if bleaching is present and correct for it. Note that 3D stacks and time series of widefield images will always be corrected and confocal only if bleaching over time is exponential.
- **Brick layout**: Option to split the images into bricks for processing in case the computers memory is not large enough to deconvolve the image as a whole. Leave it at `Auto`.
- Click `Deconvolve`.
- The results are displayed on right panel. Pressing `Accept` will cause the process to move to next channel.
- After all channels are deconvolved, we can combine the results into final deconvolved multichannel image. Again the result will be shown on right panel.
- Once finished press `Done` to exit wizard. **Press** `Save template` **to save this routine as a template to run on other images or for batch processing.**
- After exiting the wizard, the result can be seen in the main viewer window. This is not saved yet. **So save this image to your disk.**
- Tips:
- Instead of `Accept`, press `Restart` and run the process for different SNR such as 40, 60, 80. Click `All done` once finished.
- Again, combine the results from different SNR into multichannel image with original image as ch 0. Click `Next arrow`.
- Run the twin slicer to inspect the result multichannel image.
- Select the xz plane on the bottom of the twin slicer window.
- In the left panel of the window, draw a line through a bright spot along with some background to plot the signal along this line. The plot will be drawn in the right panel.
- Use the magnifying glass button on top of the plot to zoom into the diagram.
- Choose `False color` in `Channels and Colors` to see profile plots from different channels corresponding to different SNR settings in different colors.
- You will see that the signal along the line is quite different between the original image (ch 0) and the deconvolved images. Furthermore, with reaching a certain SNR, the deconvolved images do not change anymore. Choose a value such that peak is clearly visible instead of a just a small bump (as for original image). For example the images calculated using SNR 20, 40 and 80 may show a very similar curve. Thus, for deconvolution of these images an SNR configuration of at least 20 will provide good results.
This diff is collapsed.
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment