|Extended Depth of Field|
Investigators: François Aguet, Brigitte Foster
Summary: Extended depth of field (EDF) is an image-processing technique that generates a single in-focus image and a depth map from a focal series of migrographs (z-stack) of an opaque specimen. We have developed two such EDF algorithms: (1) a fusion-based technique that uses a complex wavelet transform, and (2) a reconstruction method (2.5D deconvolution) that relies on an explicit image-formation model and jointly recovers the in-focus image (texture map) and topography of the specimen.
A limited depth of field is a common problem in biological imaging with conventional light microscopy. Often, the specimen's profile covers more than the attainable depth of field. Portions of the surface of the object outside the optical plane appear defocused in the acquired image plane. This becomes worse as the magnification increases because the numerical aperture increases, too, and therefore the depth of field becomes smaller. Consequently, each acquisition will be compromised and show certain parts of the specimen in and out of focus. One common approach to image the whole specimen is to take multiple images that correspond to different object planes. When one is dealing with an opaque specimen (e.g., a stained tumor), it is possible to reconstruct an image of the surface of the object that is sharp everywhere through a fusion process referred-to as Extended Depth of Field (EDF).
We have investigated a wavelet-based algorithm for extended depth of field. The idea is to fuse the data in the wavelet domain by retaining predominant wavelet coefficients. We have compared various transforms and found that complex-valued wavelet bases did yield the best results. In addition, we have proposed a multichannel-images extension of the method that does not suffer from false colors artifacts—an important requirement for multichannel fluorescence microscopy imaging. The basic version of this algorithm is available as a
plugin for ImageJ.
The major drawback of most methods for EDF, including the one described above, is the lack of an appropriate representation of the specimenís topography, which can render the interpretation of the fusion image difficult.
To address this limitation, we have proposed a new approach that is based on an image formation model for a thick specimen, incorporating the point spread function (PSF) of the system. The reconstruction of the fusion image and topography of the specimen is stated as a joint minimization of a least-squares criterion. When the in-focus PSF has a blurring effect, or when the true in-focus position falls in-between acquisition planes, our method also acts as a deconvolution operation. This approach has produced promising results, providing a clear improvement over traditional methods, especially for the estimated topography.
Collaborations: Prof. Michael Unser, Dr. Nathalie Garin (Centre de microscopie, ISREC)
Funding: Swiss National Science Foundation
B. Forster, D. Van De Ville, J. Berent, D. Sage, M. Unser, "Extended Depth-of-Focus for Multi-Channel Microscopy Images: A Complex Wavelet Approach," Proceedings of the Second IEEE International Symposium on Biomedical Imaging: From Nano to Macro (ISBI'04), Arlington VA, USA, April 15-18, 2004, pp. 660-663.
B. Forster, D. Van De Ville, J. Berent, D. Sage, M. Unser, "Complex Wavelets for Extended Depth-of-Field: A New Method for the Fusion of Multichannel Microscopy Images," Microscopy Research and Technique, vol. 65, no. 1-2, pp. 33-42, September 2004.