Artificial Intelligence-Powered Automated Holotomographic Microscopy Enables Label-Free Quantitative Biology
Hugo Moreno, Lorenzo Archetti, Emma Gibbin, Alexandre E. Grandchamp, and Mathieu Fréchin* Nanolive SA, Lake Geneva Park, Switzerland
*
mathieu.frechin@
nanolive.ch Abstract: Holotomographic microscopy (HTM) measures the
refractive index (RI) tomograms of living cells and tissues in three dimensions. The ability to observe biological processes at high spatial and temporal resolution opens uncharted territories for cell biologists, however, current HTM devices have a limited throughput. We show here the first automated multi-well plate-compatible HTM device, the CX-A. Thanks to state-of-the-art environment control and a new type of autofocus, the CX-A can record multiple conditions in parallel over large fields of view, while its software EVE supports automated single-cell segmentation and quantification. This opens the door to new applications for HTM, from drug screening to systems biology.
Keywords: microscopy, drug research, cell biology, computer vision, artificial intelligence
Introduction Living cells are by nature transparent. Microscopy tech-
niques overcome this problem by transforming the optical properties of the sample into an observable contrast (phase contrast, differential interference contrast [DIC]) or by using fluorescent dyes. Te latter technique, however, presents major limitations such as phototoxicity and the interference of markers with the biological processes they target, while the less perturbing classic label-free techniques, such as DIC, provide images of poor contrast and resolution. In this context, holotomographic microscopy (HTM) [1]
is of great interest since it generates rich and detailed label- free images while using a very low-power light source, hence inducing no detectable phototoxicity [2]. At its core, the device is based on quantitative phase microscopy [3]. A laser diode generates a partially coherent light beam (520 nm), which is split in two to create a Mach–Zehnder interferometer setup [4]. One of the beams, the “object beam,” interacts with the sample before being collected by a 60× objective, while the sec- ond beam remains unperturbed and serves as a reference. Te interference of the two beams (object and reference) creates a hologram that is recorded on a complementary metal oxide semiconductor (CMOS) camera. Te CX-A HTM presented here (Figure 1) uses this holographic approach and combines it with rotational scanning [1] (Figure 1a). Te collection of quantitative phase information is synthesized [5–7] in order to reconstruct a full 3D refractive index (RI) tomogram. Stabil- ity of the HTM performance is ensured by continuous opti- mal calibration monitoring during an acquisition experiment, which makes it possible to accommodate sample changes such as evaporation of the mounting medium. Tanks to its rotating illumination (Figure 1a), the CX-A
allows unique characterizations of a cell population and its cellular and organelle details through their RI distribution in space and time with unmatched resolution and contrast.
24 doi:10.1017/S1551929521001139
Bringing such HTM performance to an automated setup required two challenges to be overcome. Te first challenge was to automate the HTM acquisition process while maintaining perfect device calibration and imaging focus, to ensure HTM was compatible with microscopic screening or grid scanning for the acquisition of large fields of view. Te second challenge was to create an automated image-analysis platform able to segment and quantify every single cell present in label-free images without tedious setup processes. Terefore, the CX-A combines a holotomographic micro-
scope and an epifluorescence system mounted on an auto- mated stage (Figures 1b and 1c). Te synchronization of the various parts is ensured by the CX-A soſtware, EVE. Tis soſt- ware allows the setup of acquisition protocols and offers users the ability to visualize their data. Moreover, it contains a new quantification platform called EVE Analytics (EA), which allows for the segmentation and quantitative analysis of cells acquired with HTM. EA uses artificial intelligence (AI)-aided signal preparation and advanced object detection techniques to create a seamless image-analysis experience with almost no setup time. Te CX-A is, therefore, a new tool of choice for biologists interested in drug research, systems biology, or fundamental cell biology, to mention a few areas of research. We will describe here the key features of the CX-A, in the con- text of example applications, from compound screening to cell population growth and death.
Methods Cell culture. 3T3-derived pre-adipocytes were cultured in
DMEM complemented with 10% FBS, 1% Pen/Strep, 1% L-glu- tamine, and 1% nonessential amino acids. Fiſty thousand cells were seeded for 24 h on glass bottom FluoroDishes of 25 mm and 0.17 mm thickness (World Precision Instruments Inc., Sarasota, FL). Imaging. At its core, the imaging solution is based on
quantitative phase microscopy [8]. A laser diode generates a partially coherent light beam (wavelength of λ = 520 nm: Class 1 low-power laser, sample exposure 0.2 mW/mm2
) used to illu-
minate the sample before being collected by a 60× air objective (NA = 0.8). Te holographic data are recorded on a CMOS cam- era, and individual quantitative phase information is digitally extracted and then further numerically assembled to form a 3D refractive index (RI) tomogram [8]. HTM, in combination with epifluorescence, was performed on the CX-A (Nanolive, Tolochenaz, Switzerland). Live cell imaging. Physiological conditions for live cell
imaging were maintained with a top-stage incubator (manu- factured by TOKAI HIT, Shizuoka-ken, Japan). A constant
www.microscopy-today.com • 2021 September
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80
