Deconvolution Settings

Total Iterations:
This Indicates the total number of iterations to use for processing the dataset and by default this is set to 10.

Save Interval:
This is the interval to save the intermediate iteration results. That is, an output dataset containing the image as processed “so far” is saved after every interval as specified by this field.
It must be an even divisor of the total number of iterations.

Smoothing Factor:
The numeric value associated with this control is included in the calculation of the smoothing factor used in the Good’s Roughness Penalty portion of AutoQuant’s noise suppression scheme. The value entered is applied as 2x10^(n-4) (where n is the value entered). The default value is 0 for widefield images and 5 for confocal images.

Enable GPU Processing:
When this check box is selected the deconvolution processing will be done using the Graphics Processing Unit (GPU) of your computer (if you have a supported GPU). When this box is de-selected, the processing is done using your computer’s CPU (central processing unit). The many cores and high speed of GPU processing mean that selecting this option offers a considerable speed advantage when compared to CPU processing.

Allow XY sub-voluming:
When selected, Image-Pro is permitted to divide an image set into sub-volumes in the X and Y axes. Sub-voluming increases the speed of deconvolution but creates the possibility that problematic "tiling" artifacts may be present in the deconvolved data.

If “tiling” artifacts are present in your deconvolved data, increasing the XY Overlap(px) and XY Guardband(px) settings (as described below) is recommended. If the “tiling” artifacts remain after increasing the overlap and guardband seetings, you can try disabling the Allow XY sub-voluming setting. It is recommended to enable the Allow Z sub-voluming setting if Allow XY-sub-voluming setting is disabled. Disabling both of the sub-voluming options can lead to unexpected behavior.

Allow Z sub-voluming:
When selected, Image-Pro is permitted to divide an image set into sub-volumes in the Z axis. Whilst this setting can be helpful in limited cases for especially deep samples it is not normally recommended because the Z dimension generally contains relatively very little data in comparison to the XY planes, and breaking it down would further limit the depth-axis information that is available to deconvolution.

XY Overlap (px):
This is the number of pixels by which adjacent sub-volumes within the dataset should overlap in the X and Y axes. These pixels, common to neighboring sub-volumes, aid in blending sub-volumes during reconstruction of the final image.

XY Guardband (px):
This is the amount of padding, in pixels, that is to be added to the X and Y borders of the dataset. The guardband regions are filled with data that interpolates between one image border and the opposing border (e.g., left and right edges). This creates a smoother transition in the signal between opposite sides of the image, which reduces the potential for artifacts at the edges of the image (or sub-volumes).

By default, this will be the greater of 10 or 10x the ratio of Nyquist XY spacing to the sample's XY spacing. For example, if Nyquist spacing for a given NA and wavelength is 200nm, and the sample spacing is 100nm, the default guardband will be 10(200/100)=20.

It is recommended that this value is increased if tiling artifacts appear when using the default value. In this situation, doubling the XY guardband (and subvolume overlap) value often yields more satisfactory results.

Z Overlap (px):
This is the number of pixels by which adjacent sub-volumes within the dataset should overlap in the Z axis. These pixels, common to neighboring sub-volumes, aid in blending sub-volumes during reconstruction of the final image.

This value may be increased if Z sub-voluming is enabled, and tiling artifacts appear between sub-volumes when using the default value.

Z Guardband (px):
This is the amount of padding, in pixels, that is to be added to the Z borders of the dataset. The guardband regions are filled with data that interpolates between one image border and the opposing border (i.e., front and back edges). This creates a smoother transition in the signal between opposite sides of the image, which reduces the potential for artifacts at the edges of the image (or sub-volumes).

It is recommended that this value is increased if tiling artifacts appear when using the default value. In this situation, doubling the Z guardband (and subvolume overlap) value often yields more satisfactory results.

Apply PSF Constraints:
This setting allows Image-Pro to apply a bandlimit constraint to the PSF. The effect of this is to increase the smoothing and to reduce noise in the deconvolved dataset. This setting is only applicable to, and available for confocal datasets (this setting is not displayed for modalities other than confocal).

Spherical Aberration Bias:
If you suspect that your raw data was subjected to spherical aberration during acquisition you can compensate and correct for this in the deconvolved dataset using the spherical aberration bias settings.

Sample Embedding RI:
Select the medium between the cover-glass and the slide using this drop-down menu.

Distance:
Set the distance in micrometres between the cover-glass and the sample. This value needs to be entered for all the channels in the dataset.

A preview of the theoretical PSF is displayed to the right of the Spherical Aberration bias controls.