Fast Noninvasive Morphometric Characterization of Free Human Sperms Using Deep Learning.

The selection of high-quality sperms is critical to intracytoplasmic sperm injection, which accounts for 70­80% of in vitro fertilization (IVF) treatments. So far, sperm screening is usually performed manually by clinicians. However, the performance of manual screening is limited in its objectivity, consistency, and efficiency. To overcome these limitations, we have developed a fast and noninvasive three-stage method to characterize morphology of freely swimming human sperms in bright-field microscopy images using deep learning models. Specifically, we use an object detection model to identify sperm heads, a classification model to select in-focus images, and a segmentation model to extract geometry of sperm heads and vacuoles. The models achieve an F1-score of 0.951 in sperm head detection, a z-position estimation error within ±1.5 µm in in-focus image selection, and a Dice score of 0.948 in sperm head segmentation, respectively. Customized lightweight architectures are used for the models to achieve real-time analysis of 200 frames per second. Comprehensive morphological parameters are calculated from sperm head geometry extracted by image segmentation. Overall, our method provides a reliable and efficient tool to assist clinicians in selecting high-quality sperms for successful IVF. It also demonstrates the effectiveness of deep learning in real-time analysis of live bright-field microscopy images.

Effect of a newly synthesized quinoline-based compound (PPQ-8) on murine schistosomiasis mansoni.

Schistosomiasis represents a public health problem and praziquantel is the only drug used for treatment of all forms of the disease. Thus, the development of new anti-schistosomal agents is of utmost importance to increase the effectiveness, reduce side effects and delay the emergence of resistance. The present study was conducted to report the therapeutic efficacy of PPQ-8, a new synthetic quinoline-based compound against Schistosoma mansoni. Mice were treated with PPQ-8 at day 49 post infection using two treatment regimens (20 and 40 mg/kg). Significant reductions were recorded in hepatic (62.9% and 83.6%) and intestinal tissue egg load (57.4% and 73.5%), granuloma count (75.4% and 89.1%) and diameter (26.2% and 47.3%), in response to the drug regimens, respectively. In addition, both treatment regimens induced significant decrease in liver (23.3% and 32.8%) and spleen (37.5% and 45.3%) indices. Also, there were significant reductions in mature ova, total worm and female count, which were more prominent with the higher dose. The reduction in the level of nitric oxide in the liver by both therapeutic regimens to 22.5% and 47.2% indicates the anti-oxidant activity of PPQ-8. Bright field microscopic examination of worms recovered from infected and PPQ-8-treated mice showed nearly empty intestinal caeca with no observable changes in the tegument. Our findings hold promise for the development of a novel anti-schistosomal drug using PPQ-8, but further in vitro and in vivo studies are needed to elucidate the possible mechanism/s of action and to study the effect of PPQ-8 on other human schistosomes.

Generic Isolated Cell Image Generator.

Building automated cancer screening systems based on image analysis is currently a hot topic in computer vision and medical imaging community. One of the biggest challenges of such systems, especially those using state-of-the-art deep learning techniques, is that they usually require a large amount of training data to be accurate. However, in the medical field, the confidentiality of the data and the need for medical expertise to label them significantly reduce the amount of training data available. A common practice to overcome this problem is to apply data set augmentation techniques to artificially increase the size of the training data set. Classical data set augmentation methods such as geometrical or color transformations are efficient but still produce a limited amount of new data. Hence, there has been interest in data set augmentation methods using generative models able to synthesize a wider variety of new data. VitaDX is actually developing an automated bladder cancer screening system based on the analysis of cell images contained in urinary cytology digital slides. Currently, the number of available labeled cell images is limited and therefore exploitation of the full potential of deep learning techniques is not possible. In an attempt to increase the number of labeled cell images, a new generic generator for 2D cell images has been developed and is described in this article. This framework combines previous works on cell image generation and a recent style transfer method referred to as doodle-style transfer in this article. To the best of our knowledge, we are the first to use a doodle-style transfer method for synthetic cell image generation. This framework is quite modular and could be applied to other cell image generation problems. A statistical evaluation has shown that features of real and synthetic cell images followed roughly the same distribution. Finally, the realism of the synthetic cell images has been assessed through a visual evaluation performed with the help of medical experts. © 2019 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.

Precise Pollen Grain Detection in Bright Field Microscopy Using Deep Learning Techniques.

The determination of daily concentrations of atmospheric pollen is important in the medical and biological fields. Obtaining pollen concentrations is a complex and time-consuming task for specialized personnel. The automatic location of pollen grains is a handicap due to the high complexity of the images to be processed, with polymorphic and clumped pollen grains, dust, or debris. The purpose of this study is to analyze the feasibility of implementing a reliable pollen grain detection system based on a convolutional neural network architecture, which will be used later as a critical part of an automated pollen concentration estimation system. We used a training set of 251 videos to train our system. As the videos record the process of focusing the samples, this system makes use of the 3D information presented by several focal planes. Besides, a separate set of 135 videos (containing 1234 pollen grains of 11 pollen types) was used to evaluate detection performance. The results are promising in detection (98.54% of recall and 99.75% of precision) and location accuracy (0.89 IoU as the average value). These results suggest that this technique can provide a reliable basis for the development of an automated pollen counting system.

Identification of robust focus measure functions for the automated capturing of focused images from Ziehl-Neelsen stained sputum smear microscopy slide.

Capturing of the best-focused image using focus measure function (FMF) from a stack of images acquired at different focus distances is a crucial step in automatic microscopy development. Detection of bacilli present in Ziehl-Neelsen (ZN) stained sputum smears conventional microscope (CM) images is critical to disease diagnosis. Studies have revealed that the performances of FMFs are sensitive to image contents and background noise. In this article, 24 diverse FMFs were implemented on 31 stacks of CM's field of view images acquired from three different microscopes to determine the best-focused one. Seven FMFs achieved the accuracies of greater than 90%. Accuracy, focus error, and false maxima were calculated for each FMF, and overall score and ranking were also calculated for better interpretation. Preprocessing techniques such as filtering and image distortions (noise, contrast, saturation, illumination, etc.) were performed to evaluate the robustness of every FMF. Gaussian derivative, steerable filters, Tenengrad, and Hemli and Scherer's mean FMFs were identified as the most robust and accurate functions with the accuracy >90%. These FMFs have a relatively less focus error and false maxima rate. Full widths at half maximum of these four FMFs were also computed to determine their efficacy for the optimization process. These four FMFs can be implemented for automated capturing of the image from ZN-stained sputum smear slide. Gaussian derivative FMF can also be used effectively for both CM and fluorescence microscope's field of view image stacks to determine the best-focused one from each stack. © 2017 International Society for Advancement of Cytometry.

Bright Field Microscopy to Detect Decoy Cells Due to BK Virus Infection in the Fresh and Unstained Urine Sediment in Kidney Allograft Recipients.

BACKGROUND: BK virus (BKV) may reactivate in kidney allograft recipients ultimately leading to BKV nephropathy and graft loss. Decoy cells (DCs) are one of the early marks of BKV reactivation, and these can be detected in the urine sediment. METHODS: A cohort of 102 kidney transplant patients was followed during months 3 and 6 after the transplant procedure. Urine samples were obtained to detect the presence of DC in the fresh and unstained urine sediment under bright field microscopy (BFM), in concomitance to the determination of the amount of BK viruria by qPCR. RESULTS: Decoy cells were found in 14.7% of patients (15/102). There was a strong agreement (P < 0.001) between qualitative DC detection by two experienced analysts and by qPCR. The positive predictive value, negative predictive value, specificity, and accuracy of BFM were 80%, 75%, 97%, and 75%, respectively. Test sensitivity was 16%. The comparative method was the qPCR. CONCLUSIONS: Despite its limited sensitivity, BFM of unstained urine sediment is an easily available, fast and cheap method to identify DCs in the population of kidney allograft recipients. The diagnostic performance of BFM on the hands of less experienced analysts deserves further investigation.

Simulation of bright-field microscopy images depicting pap-smear specimen.

As digital imaging is becoming a fundamental part of medical and biomedical research, the demand for computer-based evaluation using advanced image analysis is becoming an integral part of many research projects. A common problem when developing new image analysis algorithms is the need of large datasets with ground truth on which the algorithms can be tested and optimized. Generating such datasets is often tedious and introduces subjectivity and interindividual and intraindividual variations. An alternative to manually created ground-truth data is to generate synthetic images where the ground truth is known. The challenge then is to make the images sufficiently similar to the real ones to be useful in algorithm development. One of the first and most widely studied medical image analysis tasks is to automate screening for cervical cancer through Pap-smear analysis. As part of an effort to develop a new generation cervical cancer screening system, we have developed a framework for the creation of realistic synthetic bright-field microscopy images that can be used for algorithm development and benchmarking. The resulting framework has been assessed through a visual evaluation by experts with extensive experience of Pap-smear images. The results show that images produced using our described methods are realistic enough to be mistaken for real microscopy images. The developed simulation framework is very flexible and can be modified to mimic many other types of bright-field microscopy images.

Direct imaging of octahedral distortion in a complex molybdenum vanadium mixed oxide.

Complex Mo,V-based mixed oxides that crystallize in the orthorhombic M1-type structure are promising candidates for the selective oxidation of small alkanes. The oxygen sublattice of such a complex oxide has been studied by annular bright field scanning transmission electron microscopy. The recorded micrographs directly display the local distortion in the metal oxygen octahedra. From the degree of distortion we are able to draw conclusions on the distribution of oxidation states in the cation columns at different sites. The results are supported by X-ray diffraction and electron paramagnetic resonance measurements that provide integral details about the crystal structure and spin coupling, respectively.

Influence of the phase effect on gradient-based and statistics-based focus measures in bright field microscopy.

Autofocusing is essential to high throughput microscopy and live cell imaging and requires reliable focus measures. Phase objects such as separated single Chinese hamster ovary cells are almost invisible at the optical focus position in bright field microscopy images. Because of the phase effect, defocused images of phase objects have more contrast. In this paper, we show that widely used focus measures exhibit an untypical behaviour for such images. In the case of homogeneous cells, that is, when most cells tend to lie in the same focal plane, both gradient-based and statistics-based focus measures tend to have a local minimum instead of a global maximum at the optical focus position. On the other hand, if images show inhomogeneous cells, gradient-based focus measures tend to yield typical focus curves, whereas statistics-based focus measures deliver curves similar to the case of homogeneous cells. These results were interpreted using the equation describing the phase effect and patch-wise analysis of the focus curves. Bioprocess engineering experts are also influenced by the phase effect. Forty-four focus positions selected by them led to the conclusion that they prefer to look at defocused images instead of those at the optical focus.

Visualization of Individual RNA Molecules by Proximity Ligation-Based Chromogenic In Situ Hybridization Assay.

RNA in situ hybridization reveals the abundance and location of gene expression in cells or tissues, providing a technical basis for the clinical diagnosis of diseases. In this chapter, we show a "V" shape probe-mediated single-molecule chromogenic in situ hybridization (vsmCISH) technique for bright-field visualization of individual RNA molecules. In our method, several pairs of target hybridization probes are hybridized to RNA molecules and each probe pair forms a "V" shape overhang. The overhang oligonucleotides then mediated the proximity ligation to form DNA circles, followed by rolling circle amplification for signal enhancement and enzyme-catalyzed chromogenic reaction-based readout. The colorimetric assay avoids problems such as photobleaching and autofluorescence of current fluorescent in situ hybridization-based single-molecule RNA detection techniques. Furthermore, the relatively straightforward protocol makes the method useful for biological research and clinical diagnosis applications.

Fiduciary-Free Frame Alignment for Robust Time-Lapse Drift Correction Estimation in Multi-Sample Cell Microscopy.

When analyzing microscopic time-lapse observations, frame alignment is an essential task to visually understand the morphological and translation dynamics of cells and tissue. While in traditional single-sample microscopy, the region of interest (RoI) is fixed, multi-sample microscopy often uses a single microscope that scans multiple samples over a long period of time by laterally relocating the sample stage. Hence, the relocation of the optics induces a statistical RoI offset and can introduce jitter as well as drift, which results in a misaligned RoI for each sample's time-lapse observation (stage drift). We introduce a robust approach to automatically align all frames within a time-lapse observation and compensate for frame drift. In this study, we present a sub-pixel precise alignment approach based on recurrent all-pairs field transforms (RAFT); a deep network architecture for optical flow. We show that the RAFT model pre-trained on the Sintel dataset performed with near perfect precision for registration tasks on a set of ten contextually unrelated time-lapse observations containing 250 frames each. Our approach is robust for elastically undistorted and translation displaced (x,y) microscopic time-lapse observations and was tested on multiple samples with varying cell density, obtained using different devices. The approach only performed well for registration and not for tracking of the individual image components like cells and contaminants. We provide an open-source command-line application that corrects for stage drift and jitter.

High-Speed Optical Characterization of Protein-and-Nanoparticle-Stabilized Microbubbles for Ultrasound-Triggered Drug Release.

OBJECTIVE: Ultrasound-triggered bubble-mediated local drug delivery has shown potential to increase therapeutic efficacy and reduce systemic side effects, by loading drugs into the microbubble shell and triggering delivery of the payload on demand using ultrasound. Understanding the behavior of the microbubbles in response to ultrasound is crucial for efficient and controlled release. METHODS: In this work, the response of microbubbles with a coating consisting of poly(2-ethyl-butyl cyanoacrylate) (PEBCA) nanoparticles and denatured casein was characterized. High-speed recordings were taken of single microbubbles, in both bright field and fluorescence. RESULTS: The nanoparticle-loaded microbubbles show resonance behavior, but with a large variation in response, revealing a substantial interbubble variation in mechanical shell properties. The probability of shell rupture and the probability of nanoparticle release were found to strongly depend on microbubble size, and the most effective size was inversely proportional to the driving frequency. The probabilities of both rupture and release increased with increasing driving pressure amplitude. Rupture of the microbubble shell occurred after fewer cycles of ultrasound as the driving pressure amplitude or driving frequency was increased. CONCLUSION: The results highlight the importance of careful selection of the driving frequency, driving pressure amplitude and duration of ultrasound to achieve the most efficient ultrasound-triggered shell rupture and nanoparticle release of protein-and-nanoparticle-stabilized microbubbles.

Combining deep learning and droplet microfluidics for rapid and label-free antimicrobial susceptibility testing of colistin.

Efficient tools for rapid antibiotic susceptibility testing (AST) are crucial for appropriate use of antibiotics, especially colistin, which is now often considered a last resort therapy with extremely drug resistant Gram-negative bacteria. Here, we developed a rapid, easy and miniaturized colistin susceptibility assay based on microfluidics, which allows for culture and high-throughput analysis of bacterial samples. Specifically, a simple microfluidic platform that can easily be operated was designed to encapsulate bacteria in nanoliter droplets and perform a fast and automated bacterial growth detection in 2 h, using standardized samples. Direct bright-field imaging of compartmentalized samples proved to be a faster and more accurate detection method as compared to fluorescence-based analysis. A deep learning powered approach was implemented for the sensitive detection of the growth of several strains in droplets. The DropDeepL AST method (Droplet and Deep learning-based method for AST) developed here allowed the determination of the colistin susceptibility profiles of 21 fast-growing Enterobacterales (E. coli and K. pneumoniae), including clinical isolates with different resistance mechanisms, showing 100 % categorical agreement with the reference broth microdilution (BMD) method performed simultaneously. Direct AST of bacteria in urine samples on chip also provided accurate results in 2 h, without the need of complex sample preparation procedures. This method can easily be implemented in clinical microbiology laboratories, and has the potential to be adapted to a variety of antibiotics, especially for last-line antibiotics to optimize treatment of patients infected with multi-drug resistant strains.

Dataset of chicken-embryo blood cells exposed to quercetin, methyl methanesulfonate, or cadmium chloride.

Toxicological analysis of the effects of natural compounds is frequently mandated to assess their safety. In addition to more simple in vitro cellular systems, more complex biological systems can be used to evaluate toxicity. This dataset is comprised of bright-field microscopy images of chicken-embryo blood cells, a complex biological model that recapitulates several features found in human organisms, including circulation in blood stream and biodistribution to different organs. In the presented collection of blood smear images, cells were exposed to the flavonoid quercetin, and the two mutagens methyl methanesulfonate (MMS) and cadmium chloride (CdCl2). In ovo models offer a unique opportunity to investigate the effects of various substances, pathogens, or cancer treatments on developing embryos, providing valuable insights into potential risks and therapeutic strategies. In toxicology, in ovo models allow for early detection of harmful compounds and their impact on embryonic development, aiding in the assessment of environmental hazards. In immunology, these models offer a controlled system to explore the developing immune responses and the interaction between pathogens and host defenses. Additionally, in ovo models are instrumental in oncology research as they enable the study of tumor development and response to therapies in a dynamic, rapidly developing environment. Thus, these versatile models play a crucial role in advancing our understanding of complex biological processes and guiding the development of safer therapeutics and interventions. The data presented here can aid in understanding the potential toxic effects of these substances on hematopoiesis and the overall health of the developing organism. Moreover, the large dataset of blood smear images can serve as a resource for training machine learning algorithms to automatically detect and classify blood cells, provided that specific optimized conditions such as image magnification and background light are maintained for comparison. This can lead to the development of automated tools for blood cell analysis, which can be useful in research. Moreover, the data is amenable to the use as teaching and learning resource for histology and developmental biology.

Artificial Intelligence in Point-of-Care Testing.

With the projected increase in the global population, current healthcare delivery models will face severe challenges. Rural and remote areas, whether in developed or developing countries, are characterized by the same challenges: the unavailability of hospitals, lack of trained and skilled staff performing tests, and poor compliance with quality assurance protocols. Point-of-care testing using artificial intelligence (AI) is poised to be able to address these challenges. In this review, we highlight some key areas of application of AI in point-of-care testing, including lateral flow immunoassays, bright-field microscopy, and hematology, demonstrating this rapidly expanding field of laboratory medicine.

Apoptosis Detection by Quantification of Cell Debris in Bright-Field Microscopy Images.

We describe a simple protocol for quantification of apoptosis by counting subcellular debris particles in bright-field microscopy images of cell culture media samples. The only necessary equipment is a bright-field microscope (fitted with a digital camera) and a computer for image processing and analysis. The method gives comparable results to established fluorescence markers in several different applications and is, in principle, compatible with any culture format. Given its simplicity, accessibility, and inexpensive nature, the method can complement or provide an alternative to other methods for apoptosis detection.

A platform for automated and label-free monitoring of morphological features and kinetics of spheroid fusion.

Spheroids are widely applied as building blocks for biofabrication of living tissues, where they exhibit spontaneous fusion toward an integrated structure upon contact. Tissue fusion is a fundamental biological process, but due to a lack of automated monitoring systems, the in-depth characterization of this process is still limited. Therefore, a quantitative high-throughput platform was developed to semi-automatically select doublet candidates and automatically monitor their fusion kinetics. Spheroids with varying degrees of chondrogenic maturation (days 1, 7, 14, and 21) were produced from two different cell pools, and their fusion kinetics were analyzed via the following steps: (1) by applying a novel spheroid seeding approach, the background noise was decreased due to the removal of cell debris while a sufficient number of doublets were still generated. (2) The doublet candidates were semi-automatically selected, thereby reducing the time and effort spent on manual selection. This was achieved by automatic detection of the microwells and building a random forest classifier, obtaining average accuracies, sensitivities, and precisions ranging from 95.0% to 97.4%, from 51.5% to 92.0%, and from 66.7% to 83.9%, respectively. (3) A software tool was developed to automatically extract morphological features such as the doublet area, roundness, contact length, and intersphere angle. For all data sets, the segmentation procedure obtained average sensitivities and precisions ranging from 96.8% to 98.1% and from 97.7% to 98.8%, respectively. Moreover, the average relative errors for the doublet area and contact length ranged from 1.23% to 2.26% and from 2.30% to 4.66%, respectively, while the average absolute errors for the doublet roundness and intersphere angle ranged from 0.0083 to 0.0135 and from 10.70 to 13.44°, respectively. (4) The data of both cell pools were analyzed, and an exponential model was used to extract kinetic parameters from the time-series data of the doublet roundness. For both cell pools, the technology was able to characterize the fusion rate and quality in an automated manner and allowed us to demonstrate that an increased chondrogenic maturity was linked with a decreased fusion rate. The platform is also applicable to other spheroid types, enabling an increased understanding of tissue fusion. Finally, our approach to study spheroid fusion over time will aid in the design of controlled fabrication of "assembloids" and bottom-up biofabrication of living tissues using spheroids.

Following the Beat: Imaging the Valveless Pumping Function in the Early Embryonic Heart.

In vertebrates, the coordinated beat of the early heart tube drives cardiogenesis and supports embryonic growth. How the heart pumps at this valveless stage marks a fascinating problem that is of vital significance for understanding cardiac development and defects. The developing heart achieves its function at the same time as continuous and dramatic morphological changes, which in turn modify its pumping dynamics. The beauty of this muti-time-scale process also highlights its complexity that requires interdisciplinary approaches to study. High-resolution optical imaging, particularly fast, four-dimensional (4D) imaging, plays a critical role in revealing the process of pumping, instructing numerical modeling, and enabling biomechanical analyses. In this review, we aim to connect the investigation of valveless pumping mechanisms with the recent advancements in embryonic cardiodynamic imaging, facilitating interactions between these two areas of study, in hopes of encouraging and motivating innovative work to further understand the early heartbeat.

Simple formulation and characterization of double emulsion variant designed to carry three bioactive agents.

Multiple emulsions are thermodynamically stable systems that mark applications in various fields including drug delivery systems. They allow enhanced availability of drugs, greater absorption, and present reduced toxicity, among other desirable properties. In this work, we aimed to formulate a unique double emulsion (O1/W + W1/O2/W/W) with three bioactive components viz. Ocimum tenuiflorum oil, Cocos nucifera oil and crystalline Cinnamomum camphora. Three surfactants with different HLB values viz. Tween-20, Tween-80 and Triton X-100 were used for the emulsification process. The method followed was simple as compared to current methods employed for formulating multiple emulsions. Formulation was characterized using techniques of bright field microscopy, Dynamic Light Scattering (DLS), High-Resolution Transmission Electron Microscopy (HR-TEM) and Fourier-transform infrared spectroscopy (FTIR). Image processing tools were also used to characterize the formulation, which reliably cross-verified the observations from conventional characterization techniques. The potency of individual components of emulsion was compared with the prepared double emulsion model by testing the activity on two pathologically relevant bacterial strains: Fusobacterium nucleatum (FN) and Porphyromonas gingivalis (PG).

DeepHCS++: Bright-field to fluorescence microscopy image conversion using multi-task learning with adversarial losses for label-free high-content screening.

In this paper, we propose a novel microscopy image translation method for transforming a bright-field microscopy image into three different fluorescence images to observe the apoptosis, nuclei, and cytoplasm of cells, which visualize dead cells, nuclei of cells, and cytoplasm of cells, respectively. These biomarkers are commonly used in high-content drug screening to analyze drug response. The main contribution of the proposed work is the automatic generation of three fluorescence images from a conventional bright-field image; this can greatly reduce the time-consuming and laborious tissue preparation process and improve throughput of the screening process. Our proposed method uses only a single bright-field image and the corresponding fluorescence images as a set of image pairs for training an end-to-end deep convolutional neural network. By leveraging deep convolutional neural networks with a set of image pairs of bright-field and corresponding fluorescence images, our proposed method can produce synthetic fluorescence images comparable to real fluorescence microscopy images with high accuracy. Our proposed model uses multi-task learning with adversarial losses to generate more accurate and realistic microscopy images. We assess the efficacy of the proposed method using real bright-field and fluorescence microscopy image datasets from patient-driven samples of a glioblastoma, and validate the method's accuracy with various quality metrics including cell number correlation (CNC), peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), cell viability correlation (CVC), error maps, and R2 correlation.