All deconvolution algorithms for image restoration involve reiterated deconvolution operations (using Fourier Transforms) followed by a comparison of the resultant data set to the input data set. The input data set (image stack; Z-series) is always (except for the first iteration) the previous output set.

If the comparison is poor then the deconvolution operation is repeated using the results of the previous cycle and the same PSF. This cycle of * Iterations followed by generation of the comparison value* continues until the comparison value reaches a pre-determined comparison value (e.g., 0.1% difference between input and output data sets), or until the user stops the deconvolution operation.

Methods:

**Maximum Likelihood Estimation**(MLE): Uses probability to compute the comparison value based on computed theoretical noise values. This method is best for noisy WF and confocal image Z-series'.

**Iterative Constrained Tikhonov-Miller**(ICTM): Like the MLE, this is an iterative method best for low noise Confocal imaging. It is not usually appropriate for WF imaging (which, of course, what Deconvolution Microscopy is).

Example: Using SVI Huygens Pro. After loading the image stack, define the imaging parameters ("Microscope parameters) and the deconvolution method (Deconvolution Parameters). Then start the iterative process.

Quality Change Threshold is a Restoration Parameter of the Huygens Software that makes the iterative deconvolution stop when the change in the Quality Criterion between two consecutive iterations is below its value.

1) Microscope parameters: "Tell" Huygens what the imaging parameters were for a particular data set:

Step 1:

Step 2:

2) Define the parameters of deconvolution

Deconvolving via MLE

Note the iterations and the Quality number. If plotted, this factor approaches a theoretical value.