Deconvolution: Core Concepts, Algorithms and Advanced Issues

Deconvolution is a computational technique used to remove blur from images. In this presentation image formation in the microscope will be reviewed and it will be explained what deconvolution does and why deconvolution is needed as a post acquisition processing step. The concept of Point Spread Function (PSF) will be introduced. Several approaches to deconvolution and deblurring exist from simple subtractive methods to complex iterative approaches. The major approaches will be examined. It will be explained why in the presence of noise a statistical approach is required. Blind deconvolution will be introduced. The concept of Point Spread Function will be examined in detail. The effect of microscope modality, lens parameters, and specimen parameters on PSF shape will be discussed. Theoretical PSF calculation and PSF measurement using beads will be reviewed with examples. It will be explained why theoretical and measured PSFs can differ from the true system PSF. Blind deconvolution will be further examined as a tool to deal with the uncertainty of the true form of the PSF. Practical guidelines for data acquisition will be reviewed. The relationship between sampling rate and quality of deconvolution results will be explained.