Application of regularized Richardson-Lucy algorithm for deconvolution of confocal microscopy images

Area of Science:

• Microscopy
• Image Processing
• Computational Biology

Background:

• Confocal microscopy offers improved image quality over widefield microscopy by reducing out-of-focus light.
• Image deconvolution algorithms can further enhance image quality by correcting optical and data acquisition effects.
• Maximum-likelihood algorithms with regularization are practical solutions, but parameter selection remains a challenge.

Purpose of the Study:

• To develop an open-source software package for testing deconvolution algorithms.
• To find optimal regularization parameter estimates for image deconvolution.
• To implement and evaluate the Richardson-Lucy algorithm with total variation regularization.

Main Methods:

• Implementation of the Richardson-Lucy algorithm with total variation regularization in the IOCBio Microscope software.
• Development of an automatic formula to estimate the total variation regularization parameter during deconvolution.
• Generation of synthetic images based on confocal microscopy data of rat cardiomyocytes for algorithm assessment.

Main Results:

• The developed software allows for practical application of deconvolution techniques.
• An automatic method for estimating the total variation regularization parameter was derived and validated.
• An inverse relationship between the optimal regularization parameter and image peak signal-to-noise ratio was identified.
• The estimated parameter can serve as a stopping criterion for the deconvolution process.

Conclusions:

• The developed open-source software provides an effective tool for image deconvolution in microscopy.
• Automatic estimation of the total variation regularization parameter improves the deconvolution process and image quality.
• The method is effective for enhancing images from both confocal and widefield microscopes, as demonstrated with rat cardiomyocyte images.