%Aigaion2 BibTeX export from Idiap Publications
%Thursday 21 November 2024 11:44:16 AM

@INPROCEEDINGS{Marelli_ISBI2021_2021,
         author = {Marelli, Fran{\c c}ois and Liebling, Michael},
       keywords = {Light Sheet Fluorescence Microscopy, optical microscopy, Optical projection tomography},
       projects = {Idiap, COMPBIO},
          month = apr,
          title = {Optics Versus Computation: Influence of Illumination and Reconstruction Model Accuracy in Focal-Plane-Scanning Optical Projection Tomography},
      booktitle = {2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)},
           year = {2021},
          pages = {567-570},
      publisher = {IEEE},
       location = {Nice, France},
            doi = {10.1109/ISBI48211.2021.9433834},
       abstract = {Optical Projection Tomography (OPT) imaging provides isotropic resolution for samples up to a few millimeters. High resolution OPT is achieved by deconvolving Focal-Plane-Scanning (FPS-OPT) data but it requires to accurately know the system's Point Spread Function (PSF).
While the presence of noise and inaccuracies in the PSF model or parameters affects reconstruction quality, their effect is difficult to assess quantitatively in practice and the computational cost of naive simulations is prohibitively expensive.
Here, we present an efficient approach to carry out FPS-OPT simulations for a wide range of illumination geometries, including Focal-Sheet-Scanning OPT (FSS-OPT), a method using a lateral light-sheet illumination to perform FPS-OPT.
We implement a simulation framework that can accomodate large size 3D data by dividing the forward model into elements that can be efficiently processed by GPUs. 
We compare the performance of FPS-OPT and FSS-OPT on simulated data. In the presence of Poisson noise, we show that FSS-OPT outperforms FPS-OPT with deconvolution even if all model parameters are accurately known. We then validate these results on experimentally acquired data.
The availability of an efficient 3D OPT simulation framework for quantitative comparison of imaging scenarios is an essential tool for determining efficient imaging geometries to evaluate the relative benefits of computational and hardware variations.},
            pdf = {https://publications.idiap.ch/attachments/papers/2021/Marelli_ISBI2021_2021.pdf}
}