Imaging Flow Cytometer Research Articles

Single-detector multiplex imaging flow cytometry for cancer cell classification with deep learning.

Imaging flow cytometry, which combines the advantages of flow cytometry and microscopy, has emerged as a powerful tool for cell analysis in various biomedical fields such as cancer detection. In this study, we develop multiplex imaging flow cytometry (mIFC) by employing a spatial wavelength division multiplexing technique. Our mIFC can simultaneously obtain brightfield and multi-color fluorescence images of individual cells in flow, which are excited by a metal halide lamp and measured by a single detector. Statistical analysis results of multiplex imaging experiments with resolution test lens, magnification test lens, and fluorescent microspheres validate the operation of the mIFC with good imaging channel consistency and micron-scale differentiation capabilities. A deep learning method is designed for multiplex image processing that consists of three deep learning networks (U-net, very deep super resolution, and visual geometry group 19). It is demonstrated that the cluster of differentiation 24 (CD24) imaging channel is more sensitive than the brightfield, nucleus, or cancer antigen 125 (CA125) imaging channel in classifying the three types of ovarian cell lines (IOSE80 normal cell, A2780, and OVCAR3 cancer cells). An average accuracy rate of 97.1% is achieved for the classification of these three types of cells by deep learning analysis when all four imaging channels are considered. Our single-detector mIFC is promising for the development of future imaging flow cytometers and for the automatic single-cell analysis with deep learning in various biomedical fields.

Monitoring Circulating Myeloid Cells in Peritonitis with an In Vivo Imaging Flow Cytometer.

Peritonitis is a common and life-threatening inflammatory disease. Myeloid cells are elevated in the peripheral blood and contribute to peritonitis, but their circulating dynamics are not clear. In vivo flow cytometry (IVFC) is a noninvasive technique for monitoring the dynamics of circulating cells in live animals. It has been extensively used to detect circulating tumor cells, but rarely for monitoring immune cells. Here, we describe a method adapting an intravital microscope for IVFC so that we can monitor LysM-EGFP-labeled circulating myeloid cells in a tumor necrosis factor (TNF) α-induced peritonitis mouse model. Using this IVFC method, we quantified the blood flow velocity and cell concentration in circulation. We observed a significant increase in LysM-EGFP+ cells in circulation after TNFα intraperitoneal (i.p.) injection, which reached a plateau in ~20 min. Conventional cytometry analysis showed that most LysM-EGFP+ cells were neutrophils. Increasing blood neutrophils were accompanied by neutrophil recruitment to the peritoneal cavity and neutrophil emigration from the bone marrow. We then monitored neutrophil CD64 expression in vivo and found a significant increase in TNFα-induced peritonitis. We also found that CD18 blockade doubled the circulating neutrophil number in TNFα-induced peritonitis, suggesting that CD18 is critical for neutrophil recruitment in peritonitis. Overall, we demonstrate that IVFC techniques are useful for studying the circulating dynamics of immune cells during inflammatory diseases.

Imaging flow cytometric detection of del(17p) in bone marrow and circulating plasma cells in multiple myeloma.

Detection of del(17p) in myeloma is generally performed by fluorescence in situ hybridization (FISH) on a slide with analysis of up to 200 nuclei. The small cell sample analyzed makes this a low precision test. We report the utility of an automated FISH method, called "immuno-flowFISH", to detect plasma cells with adverse prognostic risk del(17p) in bone marrow and blood samples of patients with myeloma. Bone marrow (n = 31) and blood (n = 19) samples from 35 patients with myeloma were analyzed using immuno-flowFISH. Plasma cells were identified by CD38/CD138-immunophenotypic gating and assessed for the 17p locus and centromere of chromosome 17. Cells were acquired on an AMNIS ImageStreamX MkII imaging flow cytometer using INSPIRE software. Chromosome 17 abnormalities were identified in CD38/CD138-positive cells in bone marrow (6/31) and blood (4/19) samples when the percent plasma cell burden ranged from 0.03% to 100% of cells. Abnormalities could be identified in 14.5%-100% of plasma cells. The "immuno-flowFISH" imaging flow cytometric method could detect del(17p) in plasma cells in both bone marrow and blood samples of myeloma patients. This method was also able to detect gains and losses of chromosome 17, which are also of prognostic significance. The lowest levels of 0.009% (bone marrow) and 0.001% (blood) for chromosome 17 abnormalities was below the detection limit of current FISH method. This method offers potential as a new means of identifying these prognostically important chromosomal defects, even when only rare cells are present and for serial disease monitoring.

Rapid identification of bacterial isolates using microfluidic adaptive channels and multiplexed fluorescence microscopy.

We demonstrate the rapid capture, enrichment, and identification of bacterial pathogens using Adaptive Channel Bacterial Capture (ACBC) devices. Using controlled tuning of device backpressure in polydimethylsiloxane (PDMS) devices, we enable the controlled formation of capture regions capable of trapping bacteria from low cell density samples with near 100% capture efficiency. The technical demands to prepare such devices are much lower compared to conventional methods for bacterial trapping and can be achieved with simple benchtop fabrication methods. We demonstrate the capture and identification of seven species of bacteria with bacterial concentrations lower than 1000 cells per mL, including common Gram-negative and Gram-positive pathogens such as Escherichia coli and Staphylococcus aureus. We further demonstrate that species identification of the trapped bacteria can be undertaken in the order of one-hour using multiplexed 16S rRNA-FISH with identification accuracies of 70-98% with unsupervised classification methods across 7 species of bacteria. Finally, by using the bacterial capture capabilities of the ACBC chip with an ultra-rapid antimicrobial susceptibility testing method employing fluorescence imaging and convolutional neural network (CNN) classification, we demonstrate that we can use the ACBC chip as an imaging flow cytometer that can predict the antibiotic susceptibility of E. coli cells after identification.

Detection and Quantification of Cryptococcus Uptake by Phagocytic Cells Using Imaging Flow Cytometry.

Cryptococcus neoformans, the predominant etiological agent of cryptococcosis, is an encapsulated fungal pathogen found ubiquitously in the environment that causes pneumonia and life-threatening infections of the central nervous system. Following inhalation of yeasts or desiccated basidiospores into the lung alveoli, resident pulmonary phagocytic cells aid in the identification and eradication of Cryptococcus yeast through their arsenal of pattern recognition receptors (PRRs). PRRs recognize conserved pathogen-associated molecular patterns (PAMPs), such as branched mannans, β-glucans, and chitins that are the major components of the fungal cell wall. However, the key receptors/ligand interactions required for cryptococcal recognition and eventual fungal clearance have yet to be elucidated. Here we present an imaging flow cytometer (IFC) method that offers a novel quantitative cellular imaging and population statistics tool to accurately measure phagocytosis of fungal cells. It has the capacity to measure two distinct steps of phagocytosis: association/attachment and internalization in a high-throughput and quantitative manner that is difficult to achieve with other technologies. Results from these IFC studies allow for the potential to identify PRRs required for recognition, uptake, and subsequent activation of cytokine production, as well as other effector cell responses required for fungal clearance.

In-flow tomographic imaging for single cells analysis

Gold standard imaging modalities in biological field are based on fluorescence signals providing high specificity and high resolution. Recently, Fluorescence Microscopy has been combined with microfluidics to develop instrumentations called Imaging Flow Cytometers, high-throughput tools that supply bright-field, darkfield and multiple-channels fluorescence images of each single cell passing in the Field Of View (FOV). Nevertheless, Fluorescence Microscopy has some drawbacks as phototoxicity, photobleaching, expensive costs for sample preparations and also the a-priori knowledge of the tags to be used. For these reasons label-free imaging methods greatly increase in the recent years as the Quantitative Phase Imaging (QPI) technologies for microscopy. One of the optical techniques to achieve QPI is Digital Holography. DH in microscopy has several advantages such as the possibility to numerically scan the focal distance, a properties that open to the integration of DH in microfluidics. Indeed DH combined with microfluidic circuits allows to image particles or cells flowing into the FOV at different depths. Here the capabilities of label-free single-cell imaging by DH are presented and their implications on next future biomedical applications discussed. Static or in-flow configurations will be showed describing recent results and perspectives also in combination with Artificial Intelligence architectures for future applications in biomedical and clinical fields.

Artificial Intelligence-based Flow Cytometer for Real-time Algae Monitoring

Flow cytometry is a laser-based technology that rapidly detects and analyzes the chemical and physical characteristics of single cells or particles and is already well established in environmental and toxicological studies for microalgae and bacterial quantification. This study introduces an imaging Flow Cytometer (FC) system, designed specifically for the enhanced analysis of microalgae biomass populations and aggregate groups through Artificial Intelligence (AI) integration. The FC incorporates a single flow line, critical hardware components, and a trifurcated software setup. The system employs a multi-step process for counting algae units and an Artificial Neural Network (ANN) for classifying them in groups of two or four. To demonstrate its capabilities, the system was tested on its ability to capture, count, and categorize algal units, specifically the Desmodesmus sp. morphotype with high accuracy. Furthermore, the FC's capabilities were contrasted with traditional counting methods, validating its enhanced precision and efficiency against a hematocytometer. With its capability to provide rapid, accurate, and high-throughput analyses, this innovative FC paves the way for a revolutionary approach to cellular research.

Glibenclamide-Loaded Engineered Nanovectors (GNVs) Modulate Autophagy and NLRP3-Inflammasome Activation.

Activation of the NLRP3 inflammasome in response to either exogenous (PAMPs) or endogenous (DAMPs) stimuli results in the production of IL-18, caspase-1 and IL-1β. These cytokines have a beneficial role in promoting inflammation, but an excessive activation of the inflammasome and the consequent constitutive inflammatory status plays a role in human pathologies, including Alzheimer's disease (AD). Autophagic removal of NLRP3 inflammasome activators can reduce inflammasome activation and inflammation. Likewise, inflammasome signaling pathways regulate autophagy, allowing the development of inflammatory responses but preventing excessive and detrimental inflammation. Nanotechnology led to the development of liposome engineered nanovectors (NVs) that can load and carry drugs. We verified in an in vitro model of AD-associated inflammation the ability of Glibenclamide-loaded NVs (GNVs) to modulate the balance between inflammasome activation and autophagy. Human THP1dM cells were LPS-primed and oligomeric Aß-stimulated in the presence/absence of GNVs. IL-1β, IL-18 and activated caspase-1 production was evaluated by the Automated Immunoassay System (ELLA); ASC speck formation (a marker of NLRP3 activation) was analyzed by FlowSight Imaging flow-cytometer (AMNIS); the expression of autophagy targets was investigated by RT-PCR and Western blot (WB); and the modulation of autophagy-related up-stream signaling pathways and Tau phosphorylation were WB-quantified. Results showed that GNVs reduce activation of the NLRP3 inflammasome and prevent the Aß-induced phosphorylation of ERK, AKT, and p70S6 kinases, potentiating autophagic flux and counteracting Tau phosphorylation. These preliminary results support the investigation of GNVs as a possible novel strategy in disease and rehabilitation to reduce inflammasome-associated inflammation.

A Survey on Core Flow Cytometry Facilities: Instrument Maintenance, Usage, and Funding.

Flow cytometry is a powerful tool that finds applications in various fields such as immunology, molecular biology, cancer biology, virology, and infectious disease monitoring. A significant portion of the research in these disciplines is supported by flow cytometry shared resource laboratories (SRLs). There are several types of flow cytometers available for use in SRLs, including analyzers, sorters, imaging flow cytometers, and mass cytometers. Each type has different challenges when it comes to maintenance and life expectancy. An independent online survey was conducted to better understand instrument maintenance and turnover in flow cytometry SRLs. Questions regarding instrument uptime (availability), its usage, routine maintenance, and cost associated with it were addressed. The respondents also answered questions pertaining to the frequency of deep cleaning of the instrument and quality control. In addition, the survey queried about the source of funding used to purchase the instruments and possible reasons for a replacement. Presented herein are the results compiled from 146 core facilities that provide a look at the operation within a typical SRL, with the responses reflecting researchers' experiences with handling flow cytometers.

“Double Hit” Myeloma: Detection By FISH and Flow Cytometry in Circulating Myeloma Plasma Cells

“Double hit” myeloma is defined as the presence of 2 of the known high-risk genetic abnormalities in myeloma. This includes amp(1q), del(17p) and IGH translocations involving chromosomes 4, 16 and 20. Given the prognostic significance of these abnormalities, with poor overall survival, identification is crucial. Testing is traditionally by fluorescence in situ hybridization (FISH) on a single-site bone marrow sample at baseline or at specific disease timepoints. We explored whether amp(1q) and del(17p) “double hit” abnormalities could be detected in circulating myeloma plasma cells (CMPC). Aims: We assessed our novel imaging flow cytometry method that simultaneously integrates immunophenotyping and FISH. Called “immuno-flowFISH” this single cell method combines antigenic detection of neoplastic plasma cells, FISH for DNA loci of interest and imaging flow cytometry for automated assessment. Methods: Blood from 16 myeloma patients, at diagnosis or on therapy was studied. Following density gradient centrifugation, cells were incubated with CD38/CD138/CD319-AF647-conjugated antibodies to identify plasma cells, CD19-BV480 (to differentiate normal from neoplastic plasma cells) and CD3-BV510 (T-lymphocyte control). After fixation and cell membrane permeabilization, DNA was denatured (78 oC for 5 minutes). The cells were then incubated with FISH probes to 1q21 ( CKS1B) and 17p13 loci, and to the centromere of chromosome 17 (C17), and hybridized (24 hours for 37 oC). The nuclei were stained with SYTOX AADvanced and 50,000 to 500,000 cells acquired on the Amnis ® ImageStream ®X Mk II imaging flow cytometer (Cytek Biosciences). Digital images and quantitative data derived from computational algorithms (IDEAS software) were used to assess FISH signals overlying the counterstained nuclei. Results: A mean of 264,937 cells were acquired and 110,696 analysed. Cells with a neoplastic plasma cell phenotype (CD38/CD138/CD319-positive, CD19-negative) were present in all samples. The number of these CMPC ranged from 7 to 797, amounting to 0.011- 0.546% of all cells. FISH abnormalities were present in CMPC of 10 of the 16 cases (see Table). Isolated amp(1q) (three CKS1B FISH signals) was present in 4 cases, and isolated del(17p), (one 17p13 and two C17 FISH spots) in CMPC in 3 cases. “Double hit” abnormalities, with both amp(1q) and del(17p), were present in another 3 cases, with the following cytogenetic patterns (illustrated in the Figure): 1. amp(1q) and del(17p) in different cells (Case 8); 2. amp(1q) and del(17p) in the same cell (35.7% of CMPC), in addition to isolated amp(1q) (28.6% of CMPC) (Case 9); 3. amp(1q) and del(17p) in the same cells only (Case 10). All circulating myeloma plasma cells had diploid signals for the chromosome C17 centromeric probe and the CD3-positive T-cells diploid signals for all probes. Conclusion: Incorporating antigenic identification and FISH, with imaging flow cytometric analysis, can detect “double hit” amp(1q) and del(17p) in circulating myeloma plasma cells at high sensitivity (10 -5). This novel approach allowed sub-clonal interrogation, with amp(1q) and del(17p) “double hits” detected in the same (Cases 9, 10) and separate CMPC (Case 8). It could also identify sub-clonal acquisition of secondary del(17p) (Case 9), which may be a consequence of disease evolution or intra-tumoral clonal heterogeneity. FISH analysis of circulating plasma cells using a flow cytometric platform is achievable to low levels of disease through assessment of thousands of cells. It holds significant promise for sensitive holistic real-time blood surveillance to detect high-risk and sub-clonal “double hit” alterations at the single cell level. Blood monitoring for these genetic aberrations overcomes the limitations of single site marrow analysis and may provide insights into the disease trajectory, clonal evolution and prognosis.

Smartphone Imaging Flow Cytometry for High-Throughput Single-Cell Analysis.

We present a portable imaging flow cytometer comprising a smartphone, a small-footprint optical framework, and a PDMS-based microfluidic device. Flow cytometric analysis is performed in a sheathless manner via elasto-inertial focusing with a custom-written Android program, integrating a graphical user interface (GUI) that provides a high degree of user control over image acquisition. The proposed system offers two different operational modes. First, "post-processing" mode enables particle/cell sizing at throughputs of up to 67 000 particles/s. Alternatively, "real-time" mode allows for integrated cell/particle classification with machine learning at throughputs of 100 particles/s. To showcase the efficacy of our platform, polystyrene particles are accurately enumerated within heterogeneous populations using the post-processing mode. In real-time mode, an open-source machine learning algorithm is deployed within a custom-developed Android application to classify samples containing cells of similar size but with different morphologies. The flow cytometer can extract high-resolution bright-field images with a spatial resolution <700 nm using the developed machine learning-based algorithm, achieving classification accuracies of 97% and 93% for Jurkat and EL4 cells, respectively. Our results confirm that the smartphone imaging flow cytometer (sIFC) is capable of both enumerating single particles in flow and identifying morphological features with high resolution and minimal hardware.

Development of an Imaging Flow Cytometry Method for Fungal Cytological Profiling and Its Potential Application in Antifungal Drug Development.

Automated imaging techniques have been in increasing demand for the more advanced analysis and efficient characterization of cellular phenotypes. The success of the image-based profiling method hinges on assays that can rapidly and simultaneously capture a wide range of phenotypic features. We have developed an automated image acquisition method for fungal cytological profiling (FCP) using an imaging flow cytometer that can objectively measure over 250 features of a single fungal cell. Fungal cells were labeled with calcofluor white and FM4-64FX, which bind to the cell wall and lipophilic membrane, respectively. Images of single cells were analyzed using IDEAS® software. We first acquired FCPs of fungal cells treated with fluconazole, amphotericin B, and caspofungin, each with a distinct mode of action, to establish FCP databases of profiles associated with specific antifungal treatment. Once fully established, we investigated the potential application of this technique as a screening methodology to identify compounds with novel antifungal activity against Candida albicans and Cryptococcus neoformans. Altogether, we have developed a rapid, powerful, and novel image-profiling method for the phenotypic characterization of fungal cells, also with potential applications in antifungal drug development.

Single-detector dual-modality imaging flow cytometry for label-free cell analysis with machine learning

A dual-modality imaging flow cytometer (dIFC) that obtains both brightfield and light scattering images of single cells simultaneously on one detector was developed. The unique optical layout of our system with a knife-edge right-angle (KERA) prism is key for single-detector dual-modality imaging. By using our dIFC, label-free brightfield and light scattering images of single micrometer particles and cells (ovarian cancer cell lines A2780 and Hey) were obtained. Automatic classification of ovarian cell subtypes was achieved with an accuracy of 89.52% for light scattering imaging, and 70.25% for brightfield imaging by integration of dIFC with machine learning. Compared with the obtaining of brightfield and fluorescence images in conventional IFC, our dIFC is promising for label-free cellular analysis especially by adopting the light scattering image modality.

Temperature-dependent resistance to starvation of three contrasting freshwater ciliates

We investigated the temperature-dependent response to starvation of three contrasting freshwater ciliates (Ciliophora). The cyst-forming algivorous species Meseres corlissi and the bactivorous species Glaucomides bromelicola, which cannot form cysts, co-occur in the reservoirs (tanks) of tree bromeliads. The mixotrophic species Coleps spetai is common in many lakes. We hypothesized that the ciliates’ different traits and life strategies would affect their survival rates and temperature sensitivity under food depleted conditions. We measured the decline of the ciliate populations in microcosm experiments at different temperatures for several days. We used an imaging flow cytometer to size the ciliates and documented their morphological and physiological changes in response to starvation. We found that the cyst-forming species had the highest mortality rates but may endure long-term starvation by encystment. The sympatric, non-encysting species suffered the lowest mortality rates and could survive for more than three weeks without food. The mixotrophic species had intermediate mortality rates but showed the highest phenotypic plasticity in response to starvation. A significant fraction of the C. spetai population appeared unaffected by starvation, suggesting that the endosymbionts provided some resources to the host cells. The mean mortality rate per day of all three species increased with temperature by 0.09 °C−1.

Label-free multiphoton imaging flow cytometry.

Label-free imaging flow cytometry is a powerful tool for biological and medical research as it overcomes technical challenges in conventional fluorescence-based imaging flow cytometry that predominantly relies on fluorescent labeling. To date, two distinct types of label-free imaging flow cytometry have been developed, namely optofluidic time-stretch quantitative phase imaging flow cytometry and stimulated Raman scattering (SRS) imaging flow cytometry. Unfortunately, these two methods are incapable of probing some important molecules such as starch and collagen. Here, we present another type of label-free imaging flow cytometry, namely multiphoton imaging flow cytometry, for visualizing starch and collagen in live cells with high throughput. Our multiphoton imaging flow cytometer is based on nonlinear optical imaging whose image contrast is provided by two optical nonlinear effects: four-wave mixing (FWM) and second-harmonic generation (SHG). It is composed of a microfluidic chip with an acoustic focuser, a lab-made laser scanning SHG-FWM microscope, and a high-speed image acquisition circuit to simultaneously acquire FWM and SHG images of flowing cells. As a result, it acquires FWM and SHG images (100 × 100 pixels) with a spatial resolution of 500 nm and a field of view of 50 μm × 50 μm at a high event rate of four to five events per second, corresponding to a high throughput of 560-700 kb/s, where the event is defined by the passage of a cell or a cell-like particle. To show the utility of our multiphoton imaging flow cytometer, we used it to characterize Chromochloris zofingiensis (NIES-2175), a unicellular green alga that has recently attracted attention from the industrial sector for its ability to efficiently produce valuable materials for bioplastics, food, and biofuel. Our statistical image analysis found that starch was distributed at the center of the cells at the early cell cycle stage and became delocalized at the later stage. Multiphoton imaging flow cytometry is expected to be an effective tool for statistical high-content studies of biological functions and optimizing the evolution of highly productive cell strains.

Improving the Classification Accuracy in Label-Free Flow Cytometry Using Event-Based Vision and Simple Logistic Regression

Event-based cameras are novel bio-inspired vision sensors that do not follow the mechanism of traditional frame-based cameras. In the field of data acquisition, the replacement of CMOS cameras with event-based cameras has proved to enhance the accuracy of machine learning methods in situations where critical lighting conditions and rapid dynamics are paramount. In this paper, we investigate for the first time the use of extreme learning machines on data coming from event-based cameras in the context of flow cytometry. Except for the image sensor, the experimental setup is similar to a setup we used in (Lugnan et al., 2020) where we showed that a simple linear classifier can achieve around 10% error rate on background-subtracted cell frames. Here, we show that the the error rate of this simple imaging flow cytometer could be decreased to less than 2% just by making use of the capabilities of an event camera. Moreover, additional benefits like more sensitivity and efficient memory usage are gained. Finally, we suggest further possible improvements to the experimental setup used to record events from flowing micro-particles allowing for more accurate and stable cellx sorting.

Doxorubicin Loading into Milk and Mesenchymal Stem Cells’ Extracellular Vesicles as Drug Delivery Vehicles

Extracellular vesicles (EVs) have great potential as drug delivery vehicles. While mesenchymal/stromal stem cell (MSC) conditioned medium (CM) and milk are potentially safe and scalable sources of EVs for this purpose, the suitability of MSC EVs and milk EVs as drug delivery vehicles has never been compared and so was the objective of this study. Here EVs were separated from MSCs' CM and from milk and were characterised by nanoparticle tracking analysis, transmission electron microscopy, total protein quantification, and immunoblotting. An anti-cancer chemotherapeutic drug, doxorubicin (Dox), was then loaded into the EVs by one of three methods: by passive loading or by active loading by either electroporation or sonication. Dox-loaded EVs were analysed by fluorescence spectrophotometer, high-performance liquid chromatography (HPLC), and imaging flow cytometer (IFCM). Our study showed that EVs were successfully separated from the milk and MSC CM, with significantly (p < 0.001) higher yields of milk EVs/mL starting material compared to MSC EVs/mL of starting material. Using a fixed amount of EVs for each comparison, electroporation achieved significantly more Dox loading when compared to passive loading (p < 0.01). Indeed, of 250 µg of Dox made available for loading, electroporation resulted in 90.1 ± 12 µg of Dox loading into MSC EVs and 68.0 ± 10 µg of Dox loading into milk EVs, as analysed by HPLC. Interestingly, compared to the passive loading and electroporation approach, after sonication significantly fewer CD9+ EVs/mL (p < 0.001) and CD63+ EVs/mL (p < 0.001) existed, as determined by IFCM. This observation indicates that sonication, in particular, may have detrimental effects on EVs. In conclusion, EVs can be successfully separated from both MSC CM and milk, with milk being a particularly rich source. Of the three methods tested, electroporation appears to be superior for achieving maximum drug loading while not causing damage to EV surface proteins.

Real-time intelligent classification of COVID-19 and thrombosis via massive image-based analysis of platelet aggregates.

Microvascular thrombosis is a typical symptom of COVID-19 and shows similarities to thrombosis. Using a microfluidic imaging flow cytometer, we measured the blood of 181 COVID-19 samples and 101 non-COVID-19 thrombosis samples, resulting in a total of 6.3 million bright-field images. We trained a convolutional neural network to distinguish single platelets, platelet aggregates, and white blood cells and performed classical image analysis for each subpopulation individually. Based on derived single-cell features for each population, we trained machine learning models for classification between COVID-19 and non-COVID-19 thrombosis, resulting in a patient testing accuracy of 75%. This result indicates that platelet formation differs between COVID-19 and non-COVID-19 thrombosis. All analysis steps were optimized for efficiency and implemented in an easy-to-use plugin for the image viewer napari, allowing the entire analysis to be performed within seconds on mid-range computers, which could be used for real-time diagnosis.

A low-cost edge AI-chip-based system for real-time algae species classification and HAB prediction

Harmful Algal Blooms (HAB) are damaging to ecosystem functions and pose challenges to environmental and fisheries management. The key to HAB management and understanding the complex algal growth dynamics is the development of robust systems for real-time monitoring of algae populations and species. Previous algae classification studies mainly rely on the combination of an in-situ imaging flow cytometer and an off-site lab-based algae classification model such as Random Forest (RF) for the analysis of high-throughput images. An on-site AI algae monitoring system on top of an edge AI chip embedded with the proposed Algal Morphology Deep Neural Network (AMDNN) model is developed to achieve real-time algae species classification and HAB prediction. Based on a detailed examination of real-world algae images, dataset augmentation is first performed: consisting of orientation, flipping, blurring, and Resizing with Aspect ratio Preserved (RAP). The dataset augmentation is shown to significantly improve classification performance which is superior to that of the competitive RF model. And the attention heatmaps show that for relatively regular-shaped algal species (e.g., Vicicitus), the model weights the color and texture information heavily; while the shape-related features are more important for complex-shaped algae (e.g., Chaetoceros). The AMDNN is tested on a dataset of 11,250 algae images containing the 25 most common HAB classes in Hong Kong subtropical waters with 99.87% test accuracy. Based on the fast and accurate algae classification, the AI-chip-based on-site system is applied to a one-month dataset in February 2020; the predicted trends of total cell counts and targeted HAB species counts are in good agreement with observations. The proposed edge AI algae monitoring system provides a platform for the development of practical HAB early warning systems that can effectively support environmental risk and fisheries management.

Characterization of Circulating Tumor Cells Using Imaging Flow Cytometry in Liver Disease Patients

Hepatocellular carcinoma (HCC) is asymptomatic at an early stage which delays its timely diagnosis and treatment. Circulating tumor cells (CTCs), derived from a primary or secondary tumor, may help in the management of HCC. Here, we evaluate and characterize CTCs in liver disease patients. In total, 65 patients, categorized into liver cirrhosis (LC) (n=30) and HCC (n=35), were enrolled. Using ImagestreamX MkII imaging flow cytometer, CTCs were detected and characterized using biomarker expression of EpCAM, CK, AFP, CD45, and DRAQ5 in LC and HCC patients. CTCs were detected in 33/35 (94%) HCC patients and in 28/30 (93%) LC patients. In the HCC group, the number of biomarker-positive CTCs was higher in BCLC stage D when compared with others. EpCAM+CK was the most expressed biomarker on CTCs in LC versus HCC (83.3% vs. 77.14%), followed by AFP (80% vs. 65.71%), EpCAM (30% vs. 28.57%), and CK (16.6% vs. 14.28%). The EpCAM cell area was significantly associated (P value=0.031) with the CTC-positive status. The combination biomarker expression of CTCs cell area (EpCAM, CK, and AFP) performed well with the area under the curve of 0.92, high sensitivity, and specificity in detecting early-stage and AFP-negative HCC as well as in AFP-negative LC cases. Enumeration and cell area of CTCs may be used as a biomarker for early detection of HCC and guiding treatment.