Deep Learning on Chest X-ray Images to Detect and Evaluate Pneumonia Cases at the Era of COVID-19.

Coronavirus disease 2019 (COVID-19) is an infectious disease with first symptoms similar to the flu. COVID-19 appeared first in China and very quickly spreads to the rest of the world, causing then the 2019-20 coronavirus pandemic. In many cases, this disease causes pneumonia. Since pulmonary infections can be observed through radiography images, this paper investigates deep learning methods for automatically analyzing query chest X-ray images with the hope to bring precision tools to health professionals towards screening the COVID-19 and diagnosing confirmed patients. In this context, training datasets, deep learning architectures and analysis strategies have been experimented from publicly open sets of chest X-ray images. Tailored deep learning models are proposed to detect pneumonia infection cases, notably viral cases. It is assumed that viral pneumonia cases detected during an epidemic COVID-19 context have a high probability to presume COVID-19 infections. Moreover, easy-to-apply health indicators are proposed for estimating infection status and predicting patient status from the detected pneumonia cases. Experimental results show possibilities of training deep learning models over publicly open sets of chest X-ray images towards screening viral pneumonia. Chest X-ray test images of COVID-19 infected patients are successfully diagnosed through detection models retained for their performances. The efficiency of proposed health indicators is highlighted through simulated scenarios of patients presenting infections and health problems by combining real and synthetic health data.

Pick-and-Place Assembly of Single Microtubules.

Intracellular transport is affected by the filament network in the densely packed cytoplasm. Biophysical studies focusing on intracellular transport based on microtubule-kinesin system frequently use in vitro motility assays, which are performed either on individual microtubules or on random (or simple) microtubule networks. Assembling intricate networks with high flexibility requires the manipulation of 25 nm diameter microtubules individually, which can be achieved through the use of pick-and-place assembly. Although widely used to assemble tiny objects, pick-and-place is not a common practice for the manipulation of biological materials. Using the high-level handling capabilities of microelectromechanical systems (MEMS) technology, tweezers are designed and fabricated to pick and place single microtubule filaments. Repeated picking and placing cycles provide a multilayered and multidirectional microtubule network even for different surface topographies. On-demand assembly of microtubules forms crossings at desired angles for biophysical studies as well as complex networks that can be used as nanotransport systems.

A Vision Transformer-Based Framework for Knowledge Transfer From Multi-Modal to Mono-Modal Lymphoma Subtyping Models.

Determining lymphoma subtypes is a crucial step for better patient treatment targeting to potentially increase their survival chances. In this context, the existing gold standard diagnosis method, which relies on gene expression technology, is highly expensive and time-consuming, making it less accessibility. Although alternative diagnosis methods based on IHC (immunohistochemistry) technologies exist (recommended by the WHO), they still suffer from similar limitations and are less accurate. Whole Slide Image (WSI) analysis using deep learning models has shown promising potential for cancer diagnosis, that could offer cost-effective and faster alternatives to existing methods. In this work, we propose a vision transformer-based framework for distinguishing DLBCL (Diffuse Large B-Cell Lymphoma) cancer subtypes from high-resolution WSIs. To this end, we introduce a multi-modal architecture to train a classifier model from various WSI modalities. We then leverage this model through a knowledge distillation process to efficiently guide the learning of a mono-modal classifier. Our experimental study conducted on a lymphoma dataset of 157 patients shows the promising performance of our mono-modal classification model, outperforming six recent state-of-the-art methods. In addition, the power-law curve, estimated on our experimental data, suggests that with more training data from a reasonable number of additional patients, our model could achieve competitive diagnosis accuracy with IHC technologies. Furthermore, the efficiency of our framework is confirmed through an additional experimental study on an external breast cancer dataset (BCI dataset).

ProNGF promotes brain metastasis through TrkA/EphA2 induced Src activation in triple negative breast cancer cells.

BACKGROUND: Triple-Negative Breast Cancer is particularly aggressive, and its metastasis to the brain has a significant psychological impact on patients' quality of life, in addition to reducing survival. The development of brain metastases is particularly harmful in triple-negative breast cancer (TNBC). To date, the mechanisms that induce brain metastasis in TNBC are poorly understood. METHODS: Using a human blood-brain barrier (BBB) in vitro model, an in vitro 3D organotypic extracellular matrix, an ex vivo mouse brain slices co-culture and in an in vivo xenograft experiment, key step of brain metastasis were recapitulated to study TNBC behaviors. RESULTS: In this study, we demonstrated for the first time the involvement of the precursor of Nerve Growth Factor (proNGF) in the development of brain metastasis. More importantly, our results showed that proNGF acts through TrkA independent of its phosphorylation to induce brain metastasis in TNBC. In addition, we found that proNGF induces BBB transmigration through the TrkA/EphA2 signaling complex. More importantly, our results showed that combinatorial inhibition of TrkA and EphA2 decreased TBNC brain metastasis in a preclinical model. CONCLUSIONS: These disruptive findings provide new insights into the mechanisms underlying brain metastasis with proNGF as a driver of brain metastasis of TNBC and identify TrkA/EphA2 complex as a potential therapeutic target.

Enzymatic reaction in droplets manipulated with liquid dielectrophoresis.

Droplet generation and transportation for biological reactions are conducted with liquid dielectrophoresis (LDEP), forming two hundred picoliter droplets and aligning them in an open environment above the micro-machined electrodes. The generation of the dielectrophoresis signals was critically examined to actuate droplets in biological solutions without excessive Joule heating. Enzymatic reactions between ß-galactosidase and fluorescein di-ß-D-galactopyranoside were succeeded in manipulated droplets, which was confirmed by fluorescence imaging. These results allow us to propose the integration of LDEP actuation in high throughput biomolecular assays.

An Evaluation of Computational Learning-based Methods for the Segmentation of Nuclei in Cervical Cancer Cells from Microscopic Images.

BACKGROUND: The manual segmentation of cellular structures on Z-stack microscopic images is time-consuming and often inaccurate, highlighting the need to develop auto-segmentation tools to facilitate this process. OBJECTIVE: This study aimed to compare the performance of three different machine learning architectures, including random forest (RF), AdaBoost, and multi-layer perceptron (MLP), for the autosegmentation of nuclei in proliferating cervical cancer cells on Z-Stack cellular microscopy proliferation images provided by the HCS Pharma. The impact of using post-processing techniques, such as the StarDist plugin and majority voting, was also evaluated. METHODS: The RF, AdaBoost, and MLP algorithms were used to auto-segment the nuclei of cervical cancer cells on microscopic images at different Z-stack positions. Post-processing techniques were then applied to each algorithm. The performance of all algorithms was compared by an expert to globally generated ground truth by calculating the accuracy detection rate, the Dice coefficient, and the Jaccard index. RESULTS: RF achieved the best accuracy, followed by the AdaBoost and then the MLP. All algorithms achieved good pixel classifications except in regions whereby the nuclei overlapped. The majority voting and StarDist plugin improved the accuracy of the segmentation but did not resolve the nuclei overlap issue. The Z-Stack analysis revealed similar segmentation results to the Z-stack layer used to train the image. However, a worse performance was noted for segmentations performed on different Z-stack positions, which were not used to train the algorithms. CONCLUSION: All machine learning architectures provided a good segmentation of nuclei in cervical cancer cells but did not resolve the problem of overlapping nuclei and Z-stack segmentation. Further research should therefore evaluate the combined segmentation techniques and deep learning architectures to resolve these issues.

Pairing cells of different sizes in a microfluidic device for immunological synapse monitoring.

Analyzing cell-cell interaction is essential to investigate how immune cells function. Elegant designs have been demonstrated to study lymphocytes and their interaction partners. However, these devices have been targeting cells of similar dimensions. T lymphocytes are smaller, more deformable, and more sensitive to pressure than many cells. This work aims to fill the gap of a method for pairing cells with different dimensions. The developed method uses hydrodynamic flow focusing in the z-direction for on-site modulation of effective channel height to capture smaller cells as single cells. Due to immune cells' sensitivity to pressure, the proposed method provides a stable system without any change in flow conditions at the analysis area throughout experiments. Paired live cells have their activities analyzed with calcium imaging at the immunological synapse formed under a controlled environment. The method is demonstrated with primary human T lymphocytes, acute myeloid leukemia (AML) cell lines, and primary AML blasts.

Exceptional plasticity of silicon nanobridges.

The plasticity of covalently bonded materials is a subject at the forefront of materials science, bearing on a wide range of technological and fundamental aspects. However, covalent materials fracture in a brittle manner when the deformation exceeds just a few per cent. It is predicted that a macroscopically brittle material like silicon can show nanoscale plasticity. Here we report the exceptional plasticity observed in silicon nanocontacts ('nanobridges') at room temperature using a special experimental setup combining a transmission electron microscope and a microelectromechanical system. When accounting for surface diffusion, we succeeded in elongating the nanocontact into a wire-like structure, with a fivefold increase in volume, up to more than twenty times the original length. Such a large plasticity was caused by the stress-assisted diffusion and the sliding of the intergranular, amorphous-like material among the nanocrystals.

Fabricating Silicon Resonators for Analysing Biological Samples.

The adaptability of microscale devices allows microtechnologies to be used for a wide range of applications. Biology and medicine are among those fields that, in recent decades, have applied microtechnologies to achieve new and improved functionality. However, despite their ability to achieve assay sensitivities that rival or exceed conventional standards, silicon-based microelectromechanical systems remain underutilised for biological and biomedical applications. Although microelectromechanical resonators and actuators do not always exhibit optimal performance in liquid due to electrical double layer formation and high damping, these issues have been solved with some innovative fabrication processes or alternative experimental approaches. This paper focuses on several examples of silicon-based resonating devices with a brief look at their fundamental sensing elements and key fabrication steps, as well as current and potential biological/biomedical applications.

A nano-needle/microtubule composite gliding on a kinesin-coated surface for target molecule transport.

An alternative method of micro/nano-transport has been achieved by using motor proteins. Microtubules on a kinesin-coated surface have potential to act as a nano-transport system. When microtubules are used as carriers, either cargo or cargo linkers are attached on the microtubule surface. Such cargo attachments can significantly affect kinesin motion. To deal with the difficulty caused by molecular attachment to the microtubule surface, the cargo loading and transport mechanism should be separated. In this work, we propose to use micromachined needles as cargo carriers which then can be transported on microtubules. Because of the separation of needle functionalization and transport mechanism, functionalization of the needles can proceed without any effect on the microtubule structure, significantly increasing the possible types of cargo. We have fabricated silicon needles in mass numbers using a simple and effective method and have shown that the microtubule-needle composites are transported without affecting the kinesin activity.

Active transport of oil droplets along oriented microtubules by kinesin molecular motors.

We demonstrate the active transport of liquid cargos in the form of oil-in-water emulsion droplets loaded on kinesin motor proteins moving along oriented microtubules. We analyze the motility properties of the kinesin motors (velocity and run length) and find that the liquid cargo in the form of oil droplets does not alter the motor function of the kinesin molecules. This work provides a novel method for handling only a few molecules/particles encapsulated inside the oil droplets and represents a key finding for the integration of kinesin-based active transport into nanoscale lab-on-a-chip devices. We also investigate the effect of the diameter of the droplets on the motility properties of the kinesin motors. The velocity is approximately constant irrespective of the diameter of the droplets whereas we highlight a strong increase of the run length when the diameter of the droplets increases. We correlate these results with the number of kinesin motors involved in the transport process and find an excellent agreement between our experimental result and a theoretical model.

Real-time transmission electron microscope observation of gold nanoclusters diffusing into silicon at room temperature.

Gold diffusion into silicon at room temperature was observed in real time with atomic resolution. Gold nanoclusters were formed on a silicon surface by an electrical discharge between a silicon tip and a gold coated tip inside an ultrahigh-vacuum transmission electron microscope (TEM) specimen chamber. At the moment of the gold nanocluster deposition, the gold nanoclusters had a crystalline structure. The crystalline structure gradually disappeared due to the interdiffusion between silicon and gold as observed after the deposition of gold nanoclusters. The shape of the nanocluster gradually changed due to the gold diffusion into the damaged silicon. The diffusion front between silicon and gold moved toward the silicon side. From the observations of the diffusion front, the gold diffusivity at room temperature was extracted. The extracted activation energy, 0.21 eV, matched the activation energy in bulk diffusion between damaged silicon and gold. This information is useful for optimizing the hybridization between solid-state and biological nanodevices in which gold is used as an adhesive layer between the two devices.

Direct measurement of the mechanical properties of a chromatin analog and the epigenetic effects of para-sulphonato-calix[4]arene.

By means of Silicon Nano Tweezers (SNTs) the effects on the mechanical properties of λ-phage DNA during interaction with calf thymus nucleosome to form an artificial chromatin analog were measured. At a concentration of 100 nM, a nucleosome solution induced a strong stiffening effect on DNA (1.1 N m-1). This can be compared to the effects of the histone proteins, H1, H2A, H3 where no changes in the mechanical properties of DNA were observed and the complex of the H3/H4 proteins where a smaller increase in the stiffness is observed (0.2 N m-1). Para-sulphonato-calix[4]arene, SC4, known for epigenetic activity by interacting specifically with the lysine groups of histone proteins, was studied for its effect on an artificial chromatin. Using a microfluidic SNT device, SC4 was titrated against the artificial chromatin, at a concentration of 1 mM in SC4 a considerable increase in stiffness, 15 N m-1, was observed. Simultaneously optical microscopy showed a physical change in the DNA structure between the tips of the SNT device. Electronic and Atomic Force microscopy confirmed this structural re-arrangement. Negative control experiments confirmed that these mechanical and physical effects were induced neither by the acidity of SC4 nor through nonspecific interactions of SC4 on DNA.

Size and Flexibility Define the Inhibition of the H3N2 Influenza Endonuclease Enzyme by Calix[n]arenes.

Inhibition of H3N2 influenza PA endonuclease activity by a panel of anionic calix[n]arenes and ß-cyclodextrin sulfate has been studied. The joint experimental and theoretical results reveal that the larger, more flexible and highly water-soluble sulfonato-calix[n]arenes have high inhibitory activity, with para-sulfonato-calix[8]arene, SC8, having an IC50 value of 6.4 µM. Molecular docking calculations show the SC8 can interact at both the polyanion binding site and also the catalytic site of H3N2 influenza PA endonuclease.

Boyden chamber-based compartmentalized tumor spheroid culture system to implement localized anticancer drug treatment.

In anticancer drug development, it is important to simultaneously evaluate both the effect of drugs on cell proliferation and their ability to penetrate tissues. To realize such an evaluation process, here, we present a compartmentalized tumor spheroid culture system utilizing a thin membrane with a through-hole to conduct localized anticancer treatment of tumor spheroids and monitor spheroid dimensions as an indicator of cell proliferation. The system is based on a commercialized Boyden chamber plate; a through-hole was bored through a porous membrane of the chamber, and the pre-existing 0.4 µm membrane pores were filled with parylene C. A HepG2 spheroid was immobilized onto the through-hole, separating the upper and lower compartments. Fluorescein (to verify the isolation between the compartments) and tirapazamine (TPZ; to treat only the lower part of the spheroid) were added to the upper and lower compartments, respectively. Since the transportation of fluorescein was blocked during treatment, i.e., the upper and lower compartments were isolated, it was confirmed that localized TPZ treatment was successfully conducted using the developed system. The effect of localized TPZ treatment on cell proliferation was estimated by measuring the maximum horizontal cross-sectional areas in the upper and lower parts of the spheroid by microscopic observations. This system can, thus, be used to perform localized anticancer drug treatment of tumor spheroids and evaluate the effect of drugs on cell proliferation.

Humidity dependence of charge transport through DNA revealed by silicon-based nanotweezers manipulation.

The study of the electrical properties of DNA has aroused increasing interest since the last decade. So far, controversial arguments have been put forward to explain the electrical charge transport through DNA. Our experiments on DNA bundles manipulated with silicon-based actuated tweezers demonstrate undoubtedly that humidity is the main factor affecting the electrical conduction in DNA. We explain the quasi-Ohmic behavior of DNA and the exponential dependence of its conductivity with relative humidity from the adsorption of water on the DNA backbone. We propose a quantitative model that is consistent with previous studies on DNA and other materials, like porous silicon, subjected to different humidity conditions.

MEMS technology for nanobio research.

Micro- and nanotechnology have gathered 20 years of increasing research efforts. This research activity began and developed with the design and the fabrication of micro- and nanomechanisms, sensors and actuators, which range from 10nm to 100mum. More recent trends focus on the transfer of this technology know-how towards nanobiological topics and very wide range applications can be addressed. Among them, this review proposes various examples that include MEMS tweezers for molecular direct handling and characterization, single molecular characterization in femto-L chambers and dynamic microarray for cell positioning. The micromachined devices are described with bio-oriented experiences that are relevant to foresee their future contribution to drug discovery.

Developing a MEMS Device with Built-in Microfluidics for Biophysical Single Cell Characterization.

This study combines the high-throughput capabilities of microfluidics with the sensitive measurements of microelectromechanical systems (MEMS) technology to perform biophysical characterization of circulating cells for diagnostic purposes. The proposed device includes a built-in microchannel that is probed by two opposing tips performing compression and sensing separately. Mechanical displacement of the compressing tip (up to a maximum of 14 µm) and the sensing tip (with a quality factor of 8.9) are provided by two separate comb-drive actuators, and sensing is performed with a capacitive displacement sensor. The device is designed and developed for simultaneous electrical and mechanical measurements. As the device is capable of exchanging the liquid inside the channel, different solutions were tested consecutively. The performance of the device was evaluated by introducing varying concentrations of glucose (from 0.55 mM (0.1%) to 55.5 mM (10%)) and NaCl (from 0.1 mM to 10 mM) solutions in the microchannel and by monitoring changes in the mechanical and electrical properties. Moreover, we demonstrated biological sample handling by capturing single cancer cells. These results show three important capabilities of the proposed device: mechanical measurements, electrical measurements, and biological sample handling. Combined in one device, these features allow for high-throughput multi-parameter characterization of single cells.

Elucidating the mechanism of the considerable mechanical stiffening of DNA induced by the couple Zn2+/Calix[4]arene-1,3-O-diphosphorous acid.

The couple Calix[4]arene-1,3-O-diphosphorous acid (C4diP) and zinc ions (Zn2+) acts as a synergistic DNA binder. Silicon NanoTweezer (SNT) measurements show an increase in the mechanical stiffness of DNA bundles by a factor of >150, at Zn2+ to C4diP ratios above 8, as compared to Zinc alone whereas C4diP alone decreases the stiffness of DNA. Electroanalytical measurements using 3D printed devices demonstrate a progression of events in the assembly of C4diP on DNA promoted by zinc ions. A mechanism at the molecular level can be deduced in which C4diP initially coordinates to DNA by phosphate-phosphate hydrogen bonds or in the presence of Zn2+ by Zn2+ bridging coordination of the phosphate groups. Then, at high ratios of Zn2+ to C4diP, interdigitated dimerization of C4diP is followed by cross coordination of DNA strands through Zn2+/C4diP inter-strand interaction. The sum of these interactions leads to strong stiffening of the DNA bundles and increased inter-strand binding.