Invited Lecture
ADVANCES IN COMPUTATIONAL IMAGING FOR QUANTITATIVE 3D FLUORESCENCE MICROSCOPY

Many biological and biomedical applications increasingly require the ability to visualize living specimens at higher resolution and in new ways that provide information not available in the past. Computational imaging plays an important role in the advances achieved in high resolution fluorescence microscopy [1]. A major challenge in high resolution microscopy for most biomedical applications is the inherent depth variability of the imaging process, i.e., as the imaging depth increases within a sample the three-dimensional (3D) point spread function (PSF) changes [2]. This is due to the refractive index (RI) mismatch in the imaging layers [2, 3], which introduces spherical aberration (SA) in microscopy images. Although computational optical sectioning microscopy (COSM) [4] has extended the fluorescence light-microscope to a 3D imaging tool, imaging of optically thick specimens is limited due to depth-induced aberration that worsens when focusing deeper into a sample and varies spatially with sample RI variability. In traditional COSM only one 3D PSF is used in the imaging reconstruction process (aka deconvolution) [5], but when the PSF variability is significant, the use of multiple depth-variant or space-variant PSFs is necessary to reduce undesirable computation artifacts [6]. Computational complexity generally increases with the number of 3-D PSFs used in the restoration and thus, reducing the sensitivity of the 3D PSF to SA could decrease the number of PSFs needed to achieve the desired image quality.

The widespread use of COSM has motivated development of new computational methodologies to improve the restored image quality by accounting for PSF variability while reducing computational cost. In this talk I will discuss two complimentary approaches that we have been developing to address depth-induced aberrations. The first approach involves the development of new computational algorithms for 3D data processing, based on either a depth-variant [7] or a space-variant [8, 9] image formation model for a traditional microscope. The latter, accounts for sample RI variability. The second approach relies on an instrument modification for PSF engineering through wavefront encoding, which renders the microscope less sensitive to depth aberration [10] and allows the use of model-based restoration methods previously developed for COSM.

References:

1. P. M. Carlton, et al., Proceedings of the National Academy of Sciences 107, 16016-16022 (2010).
2. S. F. Gibson and F. Lanni, Journal of the Optical Society of America A 9, 154-166 (1992).
3. S. Ghosh and C. Preza, Journal of Biomedical Optics, 20(7), 075003, (2015).
4. D. A. Agard, Annual Review of Biophysics and Bioengineering 13, 191–219 (1984).
5. J.-B. Sibarita, in Microscopy Techniques, J. Rietdorf, ed., pp. 1288-1291. Springer (2005).
6. C. Preza and J. A. Conchello, Journal of the Optical Society of America A 21, 1593-1601 (2004).
7. N. Patwary and C. Preza, Biomed. Opt. Express 6(10), 3826-3841 (2015).
8. S. Ghosh and C. Preza, IEEE International Symposium on Biomedical Imaging (ISBI), 789-792 (2015).
9. S. Ghosh and C. Preza, Journal of Biomedical Optics, accepted (2016).
10. S. V. King et al., Applied Optics, 54(29), 8587-8595 (2015).

Acknowledgement:
This work is supported by the National Science Foundation (DBI awards 0844682, 0852847, 1353904).