Tuesday 13 December 2016

Friday 9.12.16


Today's event:  Light-Field imaging;
Where and when: Friday 9th December 2016, 1pm-2pm, Andy's office (Room 246b, Physics and Astronomy, Kelvin Building, University of Glasgow);
Presenters: Laura and Guillem;
Presented: [1] Levoy, Marc, et al. "Light field Microscopy." ACM Transactions on Graphics (TOG) 25.3 (2006): 924-934[2] Cohen, Noy, et al. "Enhancing the performance of the light field microscope using wavefront coding." Optics express 22.20 (2014): 24817-24839[3] Georgiev, Todor, and Andrew Lumsdaine. "Superresolution with plenoptic camera 2.0." Adobe Systems Incorporated, Tech. Rep (2009); [4]  Broxton, Michael, et al. "Wave optics theory and 3-D deconvolution for the light field microscope." Optics express 21.21 (2013): 25418-25439.
Number of attendees: 22.

Second and last ICG Journal Club event of 2016, this time with many people from the Optics group too! 


Laura was the first presenter of the day, introducing the concept of light-field imaging discussing "Light field Microscopy", Levoy et al. 2006.
In a conventional image, each point contains information about the intensity of the light coming from one point of the imaged object. Instead, a light-field image contains, for each point of the imaged scene, information about the amount of light that reaches the imaging objective from different directions. This makes it possible to change the depth at which the image is focused or create perspective views of the imaged scene (all AFTER actually recording the image), and even reconstruct 3D volumes combining different refocused version of the same recorded image:
(image from "Light field Microscopy", Levoy et al. 2006)

In order to create these light-field images, an array of lenses is added to the imaging path of the microscope:
(image from "Light field Microscopy"Levoy et al. 2006)

Nothing comes for free though, and in light-field microscopy there is always a trade off between angular and lateral resolution. In the microscope presented in this article, each small lens in the lens array produces many images of the same point of the object, each corresponding to a different incoming light direction. In this case having more (and  smaller) lenses results in a better lateral resolution but a worse angular resolution.

This first paper raised questions about how the different final images are actually extracted from the recorded light-field image, and Laura also suggested an interesting article that discusses this topic in more details: 
Prevedel, Robert, et al. "Simultaneous whole-animal 3D imaging of neuronal activity using light-field microscopy." Nature methods 11.7 (2014): 727-730. 

From there:
"[...]
Light-field deconvolution.
The volume reconstruction itself can be formulated as a tomographic inverse problem27, wherein multiple different perspectives of a 3D volume are observed and linear reconstruction methods—implemented via deconvolution—are employed for computational 3D volume reconstruction. The image formation in light-field microscopes involves diffraction from both the objective and microlenses. PSFs for the deconvolution can be computed from scalar diffraction theory28.[...]"
[27] Kak, A.C. & Slaney, M. Principles of Computerized Tomographic Imaging (Society of Industrial and Applied Mathematics, 2001). [28] Gu, M. Advanced Optical Imaging Theory (Springer, 1999).
With Guillem and the three articles he presented we went into more details understanding how light-field imaging works, and we further discussed the deconvolution needed to reconstruct images focused at different scene depths.
Combining light-field imaging with wavefront coding, for example, it is possible to make the resolution of the reconstructed images vary/decade less when changing focusing depth:

We have only had two Journal Clubs so far, but it was enough for me to notice that I don't fully understand all the discussions that go on during these events, so I must admit I'm probably missing much of what has been said and discussed, but I hope this brief summary gives you an idea of what happened, and maybe even makes you want to come along next time too! 
Anyway, next time I'd better take some notes and also avoid waiting even just a few days before updating the blog!

We also had cakes and biscuits, as there always will be :)


Special thanks to: 
- The two presenters Guillem and Laura, for whom I hadn't prepared a star but who at least got one in the pictures;
- Pavi for smiling happily at my phone and not taking part in the Journal Club attendees new favorite sport of 'let's see who hydes best' :P;
- Miguel for helping me make sure there was no cake left at the end;
- Everybody for coming along!!

No comments:

Post a Comment