We present a CNN-based quantification pipeline for the imaging and analysis of adherent cell cultures. The imaging part features two CNNs dedicated to lens-free microscopy performing accelerated holographic reconstruction and phase unwrapping. The analysis part features CNNs estimating several cellular metrics. These CNNs maps phase image into 2D quantitative representations of cell positions and measurements. The outputs images are processed by a local maxima algorithm to obtain a list of cell measurements. Here, we discuss the performance and limitations of this CNN-based quantification pipeline. The advantage is the fast processing time, i.e. the analysis of ~10.000 cells in 10 seconds.
We used a simple setup for phase and fluorescence time-lapse imaging of thousands of cells in parallel directly in the incubator. These images allow us to make a population based cell-cycle analysis comparable to flow-cytometry, but also give us an access to the analysis on the level of individual cells followed over several cell cycles.
We can observe a thousand cells in parallel. We show a FUCCI marked HeLa cell culture observed over several days directly in the standard incubator. We compare our analysis to the flow cytometer data and show that we can produce a statistically relevant time-resolved measurement.
KEYWORDS: 3D image processing, Tomography, Live cell imaging, Diffraction, 3D modeling, 3D acquisition, Beam propagation method, 3D metrology, Spatial resolution, Scattering
Intensity diffraction tomography (IDT) is a 3D phase imaging technique that enables the reconstruction of the refractive indexes (RIs) and absorption of a sample. IDT targets biological imaging in a label-free manner using the optical variation within the sample and multiple tilted imaging to reconstruct the 3D map of RIs. However, standard IDT techniques reveal several drawbacks in terms of limited field of view and feasibility of imaging living samples in time-lapse conditions. We focused on time-lapse imaging of large sample (>100µm) without the need of large NA objective or immersion oil.
The challenge created by the absence of the phase information (intensity only measurements) as well as the limited illumination angle (low NA due to low magnification) has been solved using a Beam Propagation Method (BPM) embedded inside a deep leaning framework, that we call “physical neural network”. This network layers are encoding the 3D optical representation of the sample. Besides, we included in the forward model the effect of the spherical aberration introduced by the optical interfaces, which gave a strong impact on measurements under oblique illumination in terms of 3D spatial resolution.
Using this framework, we achieved 3D reconstructions of mouse embryos (>100µm) in time-lapse conditions over 7 days, observing the intrinsic embryonic development from single cell (low-scattering sample) to the blastocyst level (highly scattering sample). Such time-lapse yields quantitative information on the development and viability of embryos in view of the sub-cellular imaging capacities. Our technology opens up novel opportunities for 3D live cell imaging of whole organoids in time-lapse.
We report a single plane illumination microscope dedicated to image “Organ-on-chip”-like biostructures in microfluidic systems. By allowing 3D fluorescence imaging inside the chip, it will permit to follow in time sample’s development at cellular scale.
Optical diffraction tomography allows retrieving the 3D refractive index in a non-invasive and label-free manner. A sample is illuminated from various angles and the intensity of the diffracted light is recorded. The light wave can be calculated layer after layer and the inverse problem is usually solved using a gradient descent based algorithm.
Here we propose a solution to solve the inverse problem using a neural network where the weights of each layer are the unknown refractive index of the object. Importantly, the matrix product between each layers is replaced by the physics of light propagation.
We designed a particularly simple, compact and robust microscope for phase and fluorescent imaging. The phase-contrast image is reconstructed from a single, approximately 100 µm defocused image with an algorithm based on a constrained optimization of Fresnel diffraction model. Fluorescence image is recorded in-focus. No mechanical movement of neither sample nor objective or any other part of the system is needed to change between the phase-contrast and fluorescence modality. The change of focus between phase (out-of-focus) and fluorescence (in-focus) imaging is achieved with chromatic aberration specifically enhanced by the optical design of our system. Our microscope is sufficiently compact (10x10x10 cm^3) to fit into a standard biological incubator. The simple and robust design reduces the vibration and the drift of the sample. The absence of motorized components makes the system robust and resistant to the humid conditions inside the biological incubator. These aspects greatly facilitate the long-time observation of cell cultures.
We can observe a thousand of cells in parallel in a single field of view (1mm^2) with resolution down to 2 µm. We show FUCCI marked HeLa cell culture observed over three days directly in the incubator. FUCCI (fluorescence ubiquitination cell-cycle indicator), is a genetically encoded, two-colour (red and green), indicator of the progression through the cell cycle: the cells in G1 phase show red fluorescence nuclei while the cells in S, G2 and M phase display green fluorescence within the nuclei.
We use phase images for segmentation and tracking of the individual cells which allows us to determine the level of fluorescence in each cell in the green and red fluorescence channel. We compare the obtained statistics with the data from flow cytometer acquired at the end of the observation. We show that we can produce a statistically relevant time-resolved measurement of a cell population while keeping access to the individual cells.
Lens-free microscopy aims at recovering sample image from diffraction measurements. The acquisitions are usually processed with an inverse problem approach to retrieve the sample image (phase and absorption). The perfect reconstruction of the sample image is however difficult to achieve. Mostly because of the lack of phase information in the recording process. Recently, deep learning has been used to circumvent this challenge. Convolutional neural networks can be applied to the reconstructed image as a single pass to improve e.g. the signal-to noise ratio or the spatial resolution. Here as an alternative, we propose to alternate between the two classes of algorithms, between the inverse problem approach and the data driven approach. In doing so we intend to improve the reconstruction results but also and importantly try to address the concerns associated with the use of deep learning, namely the generalization and hallucination problems. To demonstrate the applicability of our novel approach we choose to address the case of floating cells sample acquired by means of lens-free microscopy. This is a challenging case with a lot of phase wrapping artifacts that has never been solved using inverse problem approaches only. We demonstrate that our approach is successful in performing the phase unwrapping and that it can next be applied to a very different cell sample, namely the cultures of adherent mammalian cell lines.
Research is continuously developing imaging methods to better understand the structure and function of biological systems. In this paper, we describe our work to develop lens-free microscopy as a novel mean to observe and quantify cells in 2D and 3D cell culture conditions.
At first, we developed a lens-free video microscope based on multiple wavelength acquisitions to perform time-lapse 2D imaging of dense cell culture inside the incubator. We demonstrated that novel phase retrieval techniques enable imaging thin cell samples with high concentration (~15000 cells over a large field of view of 29.4 mm2). The experimental data can next be further analyzed with existing cell profiling and tracking algorithms. As an example, we showed that a 7 days acquisition of a culture of HeLa cells leads to a dataset featuring 2.106 cell point measurements and 104 cell cycle tracks.
Recently, we extended our work to the video-microscopy of 3D organoids culture. We showed the capability of lens-free microscopy to perform 3D+time acquisitions of 3D organoids culture. To our knowledge, our technique is the only one able to reconstruct very large volumes of 3D cell culture (~5 mm3) by phase contrast imaging. This new mean of microscopy allowed us to observe a broad range of phenomena present in 3D environments, e.g. self-organizations, displacement of large clusters, merging and interconnection over long distances (>1 mm). In addition, this 3D microscope can capture the interactions of single cells and organoids with their 3D environment, e.g. traction forces generated by large cell aggregates over long distances, up to 1.5 mm.
Overall, lens-free microscopy techniques favor ease of use and label-free experimentations as well as time-lapse acquisitions of large datasets. Importantly, we consider that these lens-free microscopy technique can thus expand the repertoire of phenomena that can be studied within 2D and 3D organoids cultures.
Lens-free microscopy aims at recovering sample image from diffraction measurements. The acquisitions are usually processed with an inverse problem approach. Recently, deep learning has been used to further improve phase retrieval results. Here, we propose to alternate iteratively between the two algorithms, to improve the reconstruction results without losing data fidelity. We validated this method for the phase image recovery of floating cells sample at large density acquired by means of lens-free microscopy. This is a challenging case with a lot of phase wrapping artefacts that has never been successfully solved using inverse problem approaches only.
We designed a simple, compact and robust microscope for phase and fluorescent imaging. No mechanical movement of neither sample nor objective or any other part of the system is needed to change between the phase-contrast and fluorescence modality. We can observe a thousand cells in parallel in a single field of view with resolution down to 2 µm. We demonstrate the system on a FUCCI marked HeLa cell culture observed over several days directly in the standard incubator. We compare the obtained statistics to the flow cytometer data and show that we can produce a statistically relevant time-resolved measurement
Flow cytometry is the main technology used in hematology analyzers. However, this technology requires bulky and complex hardware systems. Lens-free imaging is an emerging microscopy technique based on a simple and compact inline holography setup. This technique enables to image a large field-of-view (≈30mm²) leading to statistical counting (>10 000 cells) in a single-shot acquisition consistent with performances required in hematology. We report high accuracy platelet count in 54 platelet-rich plasma samples. This accuracy can be achieved through a wise choice of the illumination spectral properties and an optimized algorithmic chain dedicated to small pure phase objects.
Spontaneous Raman scattering is a reliable technique for fast identification of single bacterial cells, when spectra are acquired in laboratory conditions where bacteria growth and state are controlled. We have developed a multi-modal system combining Raman spectroscopy and darkfield imaging, aiming at analysing environmental samples, typically in the field context of biological pathogens detection. Such samples are heterogeneous, both in terms of phenotype content and environmental matrix, even after a preliminary purification step. In this paper, we report a study on the identification of Bacillus Thuriengensis (BT) mimicing pathogen bacteria, embedded in a real-world matrix: a sample of surface water enriched with environmental bacterial species. The purpose is to evaluate both the detection limit of aging BT over time and the false alarm rate, in the conditions of our experiment.
Very wide field of view imaging can be used in biology to infer statistical information on cell populations from a singleshot acquisition. In particular, for applications in hematology, fluorescence wide-field of view imaging could be an alternative to standard fluorescence flow cytometry methods; it can be useful to achieve standard blood tests, such a leukocyte count or leukocyte differential count. We will introduce two 30mm2 wide-field fluorescence imaging set-ups and compare their performances. Both systems achieve 1x magnification. The first one is based on a macro-photography lens. Although such a system optimizes the resolution and field of view, it suffers from its bulkiness. With the second system, we seek miniaturization while loosening the requirements on image quality. It is based on a lens-less approach with a fiber plate optical relay. The potential of the two systems for hematology analyzes will be illustrated with the imaging of labelled white blood cells.
Lens-Free microscopy aims at recovering an observed object such as cell cultures from its diffraction measurements. Diffraction acquisitions are processed with an inverse problem approach to recover optical path difference (OPD) images of the object. Phase unwrapping issue is solved here by using a convolutional neural network (CNN) trained on simulations. The procedure was applied successfully on a neuron cells culture video acquisition.
Phase and fluorescence are complementary contrasts that are commonly used in biology. However, the coupling of these two modalities is traditionally limited to high magnification and complex imaging systems. For statistical studies of biological populations, a large field-of-view is required. We describe a 30 mm2 field-of-view dual-modality mesoscope with a 4-μm resolution. The potential of the system to address biological questions is illustrated on white blood cell numeration in whole blood and multiwavelength imaging of the human osteosarcoma (U2-OS) cells.
We propose a simple and compact microscope combining phase imaging with fluorescence. This compact setup can be easily inserted in a standard biological incubator and allows observation of cellular cultures over several days. Phase image of the sample is reconstructed from a single, slightly (~50 μm) defocused image taken under semi-coherent illumination. Fluorescence in-focus image is recorded in epi-fluorescence geometry. The phase and fluorescence images are taken sequentially using a single CMOS camera. No mechanical movement of neither sample nor objective is required to change the imaging modality. The only change is the wavelength of illumination and excitation light for phase and fluorescence imaging, respectively. The slight defocus needed for phase imaging is achieved due to specifically introduced chromatic aberration in the imaging system.
We present dual modality time-lapse movies of cellular cultures observed over several days in physiological conditions inside an incubator. A field-of-view of 3 mm2 allows observation up to thousands of cells with micro-meter spatial resolution in quasi-simultaneous phase and fluorescence mode. We believe that the simplicity, small dimensions, ease-of-use and low cost of the system make it a useful tool for biological research
Quantitative phase imaging (QPI) allows the monitoring of adherent cell cultures continuously over long time periods and it delivers an image of the cell with pixel intensities corresponding to the optical path difference (OPD). These images can be processed to quantify several cellular features. In particular, cell OPD measurements allows the estimation of the cell dry mass, an important metric to study cell mass and growth kinetics.
If the ability of QPI to provide phase-contrast images of cells is taken for granted, the accuracy and the precision of QPI cell OPD measurements can still be questioned. Indeed, the reported QPI cell measurements have not yet been assessed with any reference method (e.g. microfluidic resonators). And there is a lack of independent experimental comparison and validation which can hinder the acceptance of QPI in the realms of live-cell mass profiling.
With the aim of filling this gap, here we compare three different methods: digital holographic microscopy, quadriwave lateral sheering interferometry and lens-free microscopy (not yet established as a QPI technique). The experimental design is based on the inter-modality comparisons of OPD measurements performed over several tens of cells. To ensure consistency, we performed OPD measurements on a fixed cell culture the same day on the same location. Importantly, the statistical analysis of these measurements allowed us to estimate the precision of QPI cell OPD measurements without any reference material. In addition, we have evaluated the influence of the post-processing steps (baseline subtraction, cell segmentation) on the precision of QPI cell measurements.
Very wide-field of view imaging can provide statistical data on large cell populations in a single acquisition. In this paper, we describe a multimodal imaging system combining brightfield, phase and fluorescence contrasts. Its greater simplicity and lower cost compared to flow cytometry make it suitable for Point-Of-Care applications. The system’s resolution was characterized on calibrated beads and resolution targets. We illustrate the potential of the single-shot imaging approach in hematology by studying the specific morphologies of white blood cell sub-types. The results suggest that very wide field of view imaging could be an alternative to flow cytometry for some applications in hematology.
We present our implementation of lens-free video microscopy setup for the monitoring of adherent cell cultures. We use a multi-wavelength LED illumination together with a dedicated holographic reconstruction algorithm that allows for an efficient removal of twin images from the reconstructed phase image for densities up to those of confluent cell cultures (>500 cells/mm2). We thereby demonstrate that lens-free video microscopy, with a large field of view (~30 mm2) can enable us to capture the images of thousands of cells simultaneously and directly inside the incubator. It is then possible to trace and quantify single cells along several cell cycles. We thus prove that lens-free microscopy is a quantitative phase imaging technique enabling estimation of several metrics at the single cell level as a function of time, for example the area, dry mass, maximum thickness, major axis length and aspect ratio of each cell. Combined with cell tracking, it is then possible to extract important parameters such as the initial cell dry mass (just after cell division), the final cell dry mass (just before cell division), the average cell growth rate, and the cell cycle duration. As an example, we discuss the monitoring of a HeLa cell cultures which provided us with a data-set featuring more than 10 000 cell cycle tracks and more than 2x106 cell morphological measurements in a single time-lapse.
We propose a three-dimensional (3D) imaging platform based on lens-free microscopy to perform multi-angle acquisitions on 3D cell cultures embedded in extracellular matrix (ECM). We developed algorithms based on the Fourier diffraction theorem to perform fully 3D reconstructions of biological samples and we adapted the lens-free microscope to incubator conditions. Here we demonstrate for the first time, 3D+time lens-free acquisitions of 3D cell culture over 8 days directly into the incubator. The 3D reconstructed volume is as large as ~5 mm3 and provides a unique way to observe in the same 3D cell culture experiment multiple cell migration strategies. Namely, in a 3D cell culture of prostate epithelial cells embedded within a Matrigel® matrix, we are able to distinguish single cell ’leaders’, migration of cell clusters, migration of large aggregates of cells, and also close-gap and large-scale branching. In addition, we observe long-scale 3D deformations of the ECM that modify the geometry of the 3D cell culture. Interestingly, we also observed the opposite, i.e. we found that large aggregates of cells may deform the ECM by generating traction forces over very long distances. In sum we put forward a novel 3D lens-free microscopy tomographic technique to study the single and collective cell migrations, the cell-to-cell interactions and the cell-to-matrix interactions.
We present a simple and compact phase imaging microscope for long-term observation of non-absorbing biological samples such as unstained cells in nutritive media. The phase image is obtained from a single defocused image taken with a standard wide-field microscope. Using a semi-coherent light source allows us to computationally re-focus image post-acquisition and recover both phase and transmission of the complex specimen. The simplicity of the system reduces both the cost and its physical size and allows a long-term observation of samples directly in a standard biological incubator. The low cost of the system can contribute to the democratization of science by allowing to perform complex long-term biological experiments to the laboratories with constrained budget. In this proceeding we present several results taken with our prototype and discuss the possibilities and limitations of our system.
We introduce a label-free technique based on lens-free microscopy to perform cell counting and cell viability assay. Without the use of any labelling, the discrimination between alive and dead cells is obtained by analyzing the cells on the basis of their holographic signature. We have assessed this novel technique by comparing the obtained results in terms of viability and cell counting against automatic optical counting and trypan blue staining as reference methods. The lensfree measurements agree very well with the reference techniques up to ~30.106 cells/ml. We found a coefficient of determination R2 of 0.99 and a slope of 1.01 for the viability measurements (N=84 CHO cell samples).
Elastic Light Scattering (ELS) is an innovative technique to identify bacterial pathogens directly on culture plates. Compelling results have already been reported for agri-food applications. Here, we have developed ELS for clinical diagnosis, starting with Staphylococcus aureus early screening. Our goal is to bring a result (positive/negative) after only 6 h of growth to fight surgical-site infections. The method starts with the acquisition of the scattering pattern arising from the interaction between a laser beam and a single bacterial colony growing on a culture medium. Then, the resulting image, considered as the bacterial species signature, is analyzed using statistical learning techniques. We present a custom optical setup able to target bacterial colonies with various sizes (30-500 microns). This system was used to collect a reference dataset of 38 strains of S. aureus and other Staphyloccocus species (5459 images) on ChromIDSAID/ MRSA bi-plates. A validation set from 20 patients has then been acquired and clinically-validated according to chromogenic enzymatic tests. The best correct-identification rate between S. aureus and S. non-aureus (94.7%) has been obtained using a support vector machine classifier trained on a combination of Fourier-Bessel moments and Local- Binary-Patterns extracted features. This statistical model applied to the validation set provided a sensitivity and a specificity of 90.0% and 56.9%, or alternatively, a positive predictive value of 47% and a negative predictive value of 93%. From a clinical point of view, the results head in the right direction and pave the way toward the WHO’s requirements for rapid, low-cost, and automated diagnosis tools.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.