Chow, D. J. X. | Search bioRxiv

bioRxiv2021

Learned deconvolution using physics priors for structured light-sheet microscopy

Deconvolution is a challenging inverse problem, particularly in techniques that employ complex engineered point-spread functions, such as microscopy with propagation-invariant beams. Here, we present a deep learning method for deconvolution that, in lieu of end-to-end training with ground truths, is trained using known physics of the imaging system. Specifically, we train a generative adversarial network with images generated with the known point-spread function of the system, and combine this with unpaired experimental data that preserves perceptual content. Our method rapidly and robustly deconvolves and superresolves microscopy images, demonstrating a two-fold improvement in image contrast to conventional deconvolution methods. In contrast to common end-to-end networks that often require 1,000-10,000s paired images, our method is experimentally unsupervised and can be trained solely on a few hundred regions of interest. We demonstrate its performance on light-sheet microscopy with propagation-invariant Airy beams, including in calibration beads, oocytes, preimplantation embryos, and excised brain tissue, as well as illustrate its utility for Bessel-beam LSM. This method aims to democratise learned methods for deconvolution, as it does not require data acquisition outwith the conventional imaging protocol.

biophysics↗

bioRxiv2020

Non-invasive, label-free optical analysis to detect aneuploidy within the inner cell mass of the preimplantation embryo

Study questionCan label-free, non-invasive optical imaging by hyperspectral microscopy discern between euploid and aneuploid cells within the inner cell mass of the mouse preimplantation embryo? Summary answerHyperspectral microscopy shows a variance in metabolic activity which enables discrimination between euploid and aneuploid cells. What is known alreadyEuploid/aneuploid mosaicism affects up to 17.3% of human blastocyst embryos with trophectoderm biopsy or spent media currently utilised to diagnose aneuploidy and mosaicism in clinical in vitro fertilisation. Based on their design, these approaches will fail to diagnose the presence or proportion of aneuploid cells within the fetal lineage (inner cell mass (ICM)) of some blastocyst embryos. Study design, size, durationThe impact of aneuploidy on cellular metabolism of primary human fibroblast cells and mouse embryos was assessed by a fluorescence microscope adapted for imaging with multiple spectral channels (hyperspectral imaging). Primary human fibroblast cells with known ploidy were subjected to hyperspectral imaging to record native cell fluorescence (euploid n= 467; aneuploid n= 969). For mouse embryos, 50-70 individual euploid and aneuploid blastomeres (8-cell stage embryo) and chimeric blastocysts (40-50 per group: euploid; aneuploid; or 1:1 and 1:3 ratio of euploid:aneuploid) were utilised for hyperspectral imaging. Participants/materials, setting, methodsTwo models were employed: (i) Primary human fibroblasts with known karyotype and (ii) a mouse model of embryo aneuploidy where mouse embryos were treated with reversine, a reversible spindle assembly checkpoint inhibitor, during the 4-to 8-cell division. Individual blastomeres were dissociated from reversine treated (aneuploid) and control (euploid) 8-cell embryos and either imaged directly or used to generate chimeric blastocysts with differing ratios of euploid:aneuploid cells. Individual blastomeres and embryos were subjected to hyperspectral imaging. Changes in cellular metabolism were determined by quantification of metabolic cofactors (inferred from their autofluorescence signature): reduced nicotinamide adenine dinucleotide (NAD(P)H), flavins with the subsequent calculation of the optical redox ratio (ORR: Flavins/[NAD(P)H + Flavins]). Mathematical algorithms were applied to extract features from the autofluorescence signals of each cell/blastomere/inner cell mass to discriminate between euploid and aneuploid. Main results and the role of chanceAn increase in the relative abundance of NAD(P)H with a decrease in flavins led to a significant reduction in the ORR for aneuploid cells in both primary human fibroblasts and individual mouse blastomeres (P < 0.05). Mathematical algorithms were able to achieve good separation between (i) euploid and aneuploid primary human fibroblast cells, (ii) euploid and aneuploid mouse blastomeres cells and (iii) euploid and aneuploid chimeric blastocysts and (iv) 1:1 and 1:3 chimeric blastocysts. The accuracy of these separations was supported by receiver operating characteristic curves with areas under the curve of 0.85, 0.99, 0.87 and 0.88, respectively. We believe that the role of chance is low as multiple cellular models (human somatic cells and mouse embryos) demonstrated a consistent shift in cellular metabolism in response to aneuploidy as well as the robust capacity of mathematical features to separate euploid and aneuploid cells in a statistically significant manner. Limitations, reasons for cautionThere would be added value in determining the degree of embryo mosaicism by sequencing the inner cell mass (ICM) of individual blastocysts to correlate with metabolic profile and level of discrimination achieved using the mathematical features approach. Wider implications of the findingsHyperspectral imaging was able to discriminate between euploid and aneuploid human fibroblasts and mouse embryos. This may lead to the development of an accurate and non-invasive optical approach to assess mosaicism within the ICM of human embryos in the absence of fluorescent tags. Study funding/competing interest(s)K.R.D. is supported by a Mid-Career Fellowship from the Hospital Research Foundation (C-MCF-58-2019). This study was funded by the Australian Research Council Centre of Excellence for Nanoscale Biophotonics (CEI40100003). The authors declare that there is no conflict of interest.

cell biology↗