Convolutional neural networks for reconstruction of undersampled optical projection tomography data applied to in vivo imaging of zebrafish

Research output: Contribution to journalArticlepeer-review

Standard

Convolutional neural networks for reconstruction of undersampled optical projection tomography data applied to in vivo imaging of zebrafish. / Davis, Samuel P X; Kumar, Sunil; Alexandrov, Yuriy; Bhargava, Ajay; da Silva Xavier, Gabriela; Rutter, Guy A; Frankel, Paul; Sahai, Erik; Flaxman, Seth; French, Paul M W; McGinty, James.

In: Journal of Biophotonics, 29.08.2019.

Research output: Contribution to journalArticlepeer-review

Harvard

APA

Davis, S. P. X., Kumar, S., Alexandrov, Y., Bhargava, A., da Silva Xavier, G., Rutter, G. A., Frankel, P., Sahai, E., Flaxman, S., French, P. M. W., & McGinty, J. (2019). Convolutional neural networks for reconstruction of undersampled optical projection tomography data applied to in vivo imaging of zebrafish. Journal of Biophotonics, [e201900128]. https://doi.org/10.1002/jbio.201900128

Vancouver

Author

Davis, Samuel P X ; Kumar, Sunil ; Alexandrov, Yuriy ; Bhargava, Ajay ; da Silva Xavier, Gabriela ; Rutter, Guy A ; Frankel, Paul ; Sahai, Erik ; Flaxman, Seth ; French, Paul M W ; McGinty, James. / Convolutional neural networks for reconstruction of undersampled optical projection tomography data applied to in vivo imaging of zebrafish. In: Journal of Biophotonics. 2019.

Bibtex

@article{1d49d67c4a744011a24350c4d51c24c5,
title = "Convolutional neural networks for reconstruction of undersampled optical projection tomography data applied to in vivo imaging of zebrafish",
abstract = "Optical projection tomography (OPT) is a 3D mesoscopic imaging modality that can utilize absorption or fluorescence contrast. 3D images can be rapidly reconstructed from tomographic data sets sampled with sufficient numbers of projection angles using the Radon transform, as is typically implemented with optically cleared samples of the mm-to-cm scale. For in vivo imaging, considerations of phototoxicity and the need to maintain animals under anesthesia typically preclude the acquisition of OPT data at a sufficient number of angles to avoid artifacts in the reconstructed images. For sparse samples, this can be addressed with iterative algorithms to reconstruct 3D images from undersampled OPT data, but the data processing times present a significant challenge for studies imaging multiple animals. We show here that convolutional neural networks (CNN) can be used in place of iterative algorithms to remove artifacts-reducing processing time for an undersampled in vivo zebrafish dataset from 77 to 15 minutes. We also show that using CNN produces reconstructions of equivalent quality to compressed sensing with 40% fewer projections. We further show that diverse training data classes, for example, ex vivo mouse tissue data, can be used for CNN-based reconstructions of OPT data of other species including live zebrafish.",
keywords = "neural networks, optical tomography, preclinical imaging",
author = "Davis, {Samuel P X} and Sunil Kumar and Yuriy Alexandrov and Ajay Bhargava and {da Silva Xavier}, Gabriela and Rutter, {Guy A} and Paul Frankel and Erik Sahai and Seth Flaxman and French, {Paul M W} and James McGinty",
note = "{\textcopyright} 2019 The Authors. Journal of Biophotonics published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.",
year = "2019",
month = aug,
day = "29",
doi = "10.1002/jbio.201900128",
language = "English",
journal = "Journal of Biophotonics",
issn = "1864-063X",
publisher = "Wiley - V C H Verlag GmbH & Co. KGaA",

}

RIS

TY - JOUR

T1 - Convolutional neural networks for reconstruction of undersampled optical projection tomography data applied to in vivo imaging of zebrafish

AU - Davis, Samuel P X

AU - Kumar, Sunil

AU - Alexandrov, Yuriy

AU - Bhargava, Ajay

AU - da Silva Xavier, Gabriela

AU - Rutter, Guy A

AU - Frankel, Paul

AU - Sahai, Erik

AU - Flaxman, Seth

AU - French, Paul M W

AU - McGinty, James

N1 - © 2019 The Authors. Journal of Biophotonics published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

PY - 2019/8/29

Y1 - 2019/8/29

N2 - Optical projection tomography (OPT) is a 3D mesoscopic imaging modality that can utilize absorption or fluorescence contrast. 3D images can be rapidly reconstructed from tomographic data sets sampled with sufficient numbers of projection angles using the Radon transform, as is typically implemented with optically cleared samples of the mm-to-cm scale. For in vivo imaging, considerations of phototoxicity and the need to maintain animals under anesthesia typically preclude the acquisition of OPT data at a sufficient number of angles to avoid artifacts in the reconstructed images. For sparse samples, this can be addressed with iterative algorithms to reconstruct 3D images from undersampled OPT data, but the data processing times present a significant challenge for studies imaging multiple animals. We show here that convolutional neural networks (CNN) can be used in place of iterative algorithms to remove artifacts-reducing processing time for an undersampled in vivo zebrafish dataset from 77 to 15 minutes. We also show that using CNN produces reconstructions of equivalent quality to compressed sensing with 40% fewer projections. We further show that diverse training data classes, for example, ex vivo mouse tissue data, can be used for CNN-based reconstructions of OPT data of other species including live zebrafish.

AB - Optical projection tomography (OPT) is a 3D mesoscopic imaging modality that can utilize absorption or fluorescence contrast. 3D images can be rapidly reconstructed from tomographic data sets sampled with sufficient numbers of projection angles using the Radon transform, as is typically implemented with optically cleared samples of the mm-to-cm scale. For in vivo imaging, considerations of phototoxicity and the need to maintain animals under anesthesia typically preclude the acquisition of OPT data at a sufficient number of angles to avoid artifacts in the reconstructed images. For sparse samples, this can be addressed with iterative algorithms to reconstruct 3D images from undersampled OPT data, but the data processing times present a significant challenge for studies imaging multiple animals. We show here that convolutional neural networks (CNN) can be used in place of iterative algorithms to remove artifacts-reducing processing time for an undersampled in vivo zebrafish dataset from 77 to 15 minutes. We also show that using CNN produces reconstructions of equivalent quality to compressed sensing with 40% fewer projections. We further show that diverse training data classes, for example, ex vivo mouse tissue data, can be used for CNN-based reconstructions of OPT data of other species including live zebrafish.

KW - neural networks

KW - optical tomography

KW - preclinical imaging

U2 - 10.1002/jbio.201900128

DO - 10.1002/jbio.201900128

M3 - Article

C2 - 31386281

JO - Journal of Biophotonics

JF - Journal of Biophotonics

SN - 1864-063X

M1 - e201900128

ER -