Assessing red blood cell deformability from microscopy images using deep learning

Critical review of single-cell mechanotyping approaches for biomedical applications

Accurate mechanical measurements of cells has the potential to improve diagnostics, therapeutics and advance understanding of disease mechanisms, where high-resolution mechanical information can be measured by deforming individual cells. Here we evaluate recently developed techniques for measuring cell-scale stiffness properties; while many such techniques have been developed, much of the work examining single-cell stiffness is impacted by difficulties in standardization and comparability, giving rise to large variations in reported mechanical moduli. We highlight the role of underlying mechanical theories driving this variability, and note opportunities to develop novel mechanotyping devices and theoretical models that facilitate convenient and accurate mechanical characterisation. Moreover, many high-throughput approaches are confounded by factors including cell size, surface friction, natural population heterogeneity and convolution of elastic and viscous contributions to cell deformability. We nevertheless identify key approaches based on deformability cytometry as a promising direction for further development, where both high-throughput and accurate single-cell resolutions can be realized.

Cellular nucleus image-based smarter microscope system for single cell analysis

Cell imaging technology is undoubtedly a powerful tool for studying single-cell heterogeneity due to its non-invasive and visual advantages. It covers microscope hardware, software, and image analysis techniques, which are hindered by low throughput owing to abundant hands-on time and expertise. Herein, a cellular nucleus image-based smarter microscope system for single-cell analysis is reported to achieve high-throughput analysis and high-content detection of cells. By combining the hardware of an automatic fluorescence microscope and multi-object recognition/acquisition software, we have achieved more advanced process automation with the assistance of Robotic Process Automation (RPA), which realizes a high-throughput collection of single-cell images. Automated acquisition of single-cell images has benefits beyond ease and throughout and can lead to uniform standard and higher quality images. We further constructed a single-cell image database-based convolutional neural network (Efficient Convolutional Neural Network, E-CNN) exceeding 20618 single-cell nucleus images. Computational analysis of large and complex data sets enhances the content and efficiency of single-cell analysis with the assistance of Artificial Intelligence (AI), which breaks through the super-resolution microscope's hardware limitation, such as specialized light sources with specific wavelengths, advanced optical components, and high-performance graphics cards. Our system can identify single-cell nucleus images that cannot be artificially distinguished with an accuracy of 95.3%. Overall, we build an ordinary microscope into a high-throughput analysis and high-content smarter microscope system, making it a candidate tool for Imaging cytology.

The interaction between particles and vascular endothelium in blood flow

Particle-based drug delivery systems have shown promising application potential to treat human diseases; however, an incomplete understanding of their interactions with vascular endothelium in blood flow prevents their inclusion into mainstream clinical applications. The flow performance of nano/micro-sized particles in the blood are disturbed by many external/internal factors, including blood constituents, particle properties, and endothelium bioactivities, affecting the fate of particles in vivo and therapeutic effects for diseases. This review highlights how the blood constituents, hemodynamic environment and particle properties influence the interactions and particle activities in vivo. Moreover, we briefly summarized the structure and functions of endothelium and simulated devices for studying particle performance under blood flow conditions. Finally, based on particle-endothelium interactions, we propose future opportunities for novel therapeutic strategies and provide solutions to challenges in particle delivery systems for accelerating their clinical translation. This review helps provoke an increasing in-depth understanding of particle-endothelium interactions and inspires more strategies that may benefit the development of particle medicine.

Recent advances in deformation-assisted microfluidic cell sorting technologies

Cell sorting is an essential prerequisite for cell research and has great value in life science and clinical studies. Among the many microfluidic cell sorting technologies, label-free methods based on the size of different cell types have been widely studied. However, the heterogeneity in size for cells of the same type and the inevitable size overlap between different types of cells would result in performance degradation in size-based sorting. To tackle such challenges, deformation-assisted technologies are receiving more attention recently. Cell deformability is an inherent biophysical marker of cells that reflects the changes in their internal structures and physiological states. It provides additional dimensional information for cell sorting besides size. Therefore, in this review, we summarize the recent advances in deformation-assisted microfluidic cell sorting technologies. According to how the deformability is characterized and the form in which the force acts, the technologies can be divided into two categories: (1) the indirect category including transit-time-based and image-based methods, and (2) the direct category including microstructure-based and hydrodynamics-based methods. Finally, the separation performance and the application scenarios of each method, the existing challenges and future outlook are discussed. Deformation-assisted microfluidic cell sorting technologies are expected to realize greater potential in the label-free analysis of cells.

AI based image analysis of red blood cells in oscillating microchannels

The flow dynamics of red blood cells in vivo in blood capillaries and in vitro in microfluidic channels is complex. Cells can obtain different shapes such as discoid, parachute, slipper-like shapes and various intermediate states depending on flow conditions and their viscoelastic properties. We use artificial intelligence based analysis of red blood cells (RBCs) in an oscillating microchannel to distinguish healthy red blood cells from red blood cells treated with formaldehyde to chemically modify their viscoelastic behavior. We used TensorFlow to train and validate a deep learning model and achieved a testing accuracy of over 97%. This method is a first step to a non-invasive, label-free characterization of diseased red blood cells and will be useful for diagnostic purposes in haematology labs. This method provides quantitative data on the number of affected cells based on single cell classification.

Microsystem Advances through Integration with Artificial Intelligence

Microfluidics is a rapidly growing discipline that involves studying and manipulating fluids at reduced length scale and volume, typically on the scale of micro- or nanoliters. Under the reduced length scale and larger surface-to-volume ratio, advantages of low reagent consumption, faster reaction kinetics, and more compact systems are evident in microfluidics. However, miniaturization of microfluidic chips and systems introduces challenges of stricter tolerances in designing and controlling them for interdisciplinary applications. Recent advances in artificial intelligence (AI) have brought innovation to microfluidics from design, simulation, automation, and optimization to bioanalysis and data analytics. In microfluidics, the Navier-Stokes equations, which are partial differential equations describing viscous fluid motion that in complete form are known to not have a general analytical solution, can be simplified and have fair performance through numerical approximation due to low inertia and laminar flow. Approximation using neural networks trained by rules of physical knowledge introduces a new possibility to predict the physicochemical nature. The combination of microfluidics and automation can produce large amounts of data, where features and patterns that are difficult to discern by a human can be extracted by machine learning. Therefore, integration with AI introduces the potential to revolutionize the microfluidic workflow by enabling the precision control and automation of data analysis. Deployment of smart microfluidics may be tremendously beneficial in various applications in the future, including high-throughput drug discovery, rapid point-of-care-testing (POCT), and personalized medicine. In this review, we summarize key microfluidic advances integrated with AI and discuss the outlook and possibilities of combining AI and microfluidics.

Optofluidic imaging meets deep learning: from merging to emerging

Propelled by the striking advances in optical microscopy and deep learning (DL), the role of imaging in lab-on-a-chip has dramatically been transformed from a silo inspection tool to a quantitative "smart" engine. A suite of advanced optical microscopes now enables imaging over a range of spatial scales (from molecules to organisms) and temporal window (from microseconds to hours). On the other hand, the staggering diversity of DL algorithms has revolutionized image processing and analysis at the scale and complexity that were once inconceivable. Recognizing these exciting but overwhelming developments, we provide a timely review of their latest trends in the context of lab-on-a-chip imaging, or coined optofluidic imaging. More importantly, here we discuss the strengths and caveats of how to adopt, reinvent, and integrate these imaging techniques and DL algorithms in order to tailor different lab-on-a-chip applications. In particular, we highlight three areas where the latest advances in lab-on-a-chip imaging and DL can form unique synergisms: image formation, image analytics and intelligent image-guided autonomous lab-on-a-chip. Despite the on-going challenges, we anticipate that they will represent the next frontiers in lab-on-a-chip imaging that will spearhead new capabilities in advancing analytical chemistry research, accelerating biological discovery, and empowering new intelligent clinical applications.

Label-free microfluidic cell sorting and detection for rapid blood analysis

Blood tests are considered as standard clinical procedures to screen for markers of diseases and health conditions. However, the complex cellular background (>99.9% RBCs) and biomolecular composition often pose significant technical challenges for accurate blood analysis. An emerging approach for point-of-care blood diagnostics is utilizing "label-free" microfluidic technologies that rely on intrinsic cell properties for blood fractionation and disease detection without any antibody binding. A growing body of clinical evidence has also reported that cellular dysfunction and their biophysical phenotypes are complementary to standard hematoanalyzer analysis (complete blood count) and can provide a more comprehensive health profiling. In this review, we will summarize recent advances in microfluidic label-free separation of different blood cell components including circulating tumor cells, leukocytes, platelets and nanoscale extracellular vesicles. Label-free single cell analysis of intrinsic cell morphology, spectrochemical properties, dielectric parameters and biophysical characteristics as novel blood-based biomarkers will also be presented. Next, we will highlight research efforts that combine label-free microfluidics with machine learning approaches to enhance detection sensitivity and specificity in clinical studies, as well as innovative microfluidic solutions which are capable of fully integrated and label-free blood cell sorting and analysis. Lastly, we will envisage the current challenges and future outlook of label-free microfluidics platforms for high throughput multi-dimensional blood cell analysis to identify non-traditional circulating biomarkers for clinical diagnostics.

Data for assessing red blood cell deformability from microscopy images using deep learning

Red blood cell (RBC) deformability is a vital biophysical property that dictates the ability of these cells to repeatedly squeeze through small capillaries in the microvasculature. This capability is known to differ between individuals and degrades due to natural aging, pathology, and cold storage. There is great interest in measuring RBC deformability because this parameter is a potential biomarker of RBC quality for use in blood transfusions. Measuring this property from microscopy images would greatly reduce the effort required to acquire this information, as well as improve standardization across different centers. This dataset consists of live cell microscopy images of RBC samples from 10 healthy donors. Each RBC sample is sorted into fractions based on deformability using the microfluidic ratchet device. Each deformability fraction is imaged in microwell plates using a Nikon CFI S Plan Fluor ELWD 40 × objective and a Nikon DS-Qi2 CMOS camera on a Nikon Ti-2E inverted microscope. This data could be reused to develop deep learning algorithms to associate live cell images with cell deformability.

Predicting willingness to donate blood based on machine learning: two blood donor recruitments during COVID-19 outbreaks

Machine learning methods are a novel way to predict and rank donors' willingness to donate blood and to achieve precision recruitment, which can improve the recruitment efficiency and meet the challenge of blood shortage. We collected information about experienced blood donors via short message service (SMS) recruitment and developed 7 machine learning-based recruitment models using PyCharm-Python Environment and 13 features which were described as a method for ranking and predicting donors' intentions to donate blood with a floating number between 0 and 1. Performance of the prediction models was assessed by the Area under the receiver operating characteristic curve (AUC), accuracy, precision, recall, and F1 score in the full dataset, and by the accuracy in the four sub-datasets. The developed models were applied to prospective validations of recruiting experienced blood donors during two COVID-19 pandemics, while the routine method was used as a control. Overall, a total of 95,476 recruitments via SMS and their donation results were enrolled in our modelling study. The strongest predictor features for the donation of experienced donors were blood donation interval, age, and donation frequency. Among the seven baseline models, the eXtreme Gradient Boosting (XGBoost) and Support vector machine models (SVM) achieved the best performance: mean (95%CI) with the highest AUC: 0.809 (0.806-0.811), accuracy: 0.815 (0.812-0.818), precision: 0.840 (0.835-0.845), and F1 score of XGBoost: 0.843 (0.840-0.845) and recall of SVM: 0.991 (0.988-0.994). The hit rate of the XGBoost model alone and the combined XGBoost and SVM models were 1.25 and 1.80 times higher than that of the conventional method as a control in 2 recruitments respectively, and the hit rate of the high willingness to donate group was 1.96 times higher than that of the low willingness to donate group. Our results suggested that the machine learning models could predict and determine the experienced donors with a strong willingness to donate blood by a ranking score based on personalized donation data and demographical details, significantly improve the recruitment rate of blood donors and help blood agencies to maintain the blood supply in emergencies.

Computational Study of Methods for Determining the Elasticity of Red Blood Cells Using Machine Learning

RBC (Red Blood Cell) membrane is a highly elastic structure, and proper modelling of this elasticity is essential for biomedical applications that involve computational experiments with blood flow. In this work, we present a new method for estimating one of the key parameters of red blood cell elasticity, which uses a neural network trained on the simulation outputs. We test classic LSTM (Long-Short Term Memory) architecture for the time series regression task, and we also experiment with novel CNN-LSTM (Convolutional Neural Network) architecture. We paid special attention to investigating the impact of the way the three-dimensional training data are reduced to their two-dimensional projections. Such a comparison is possible thanks to working with simulation outputs that are equivalently defined for all dimensions and their combinations. The obtained results can be used as recommendations for an appropriate way to record real experiments for which the reduced dimension of the acquired data is essential.

Application of machine learning in predicting blood flow and red cell distribution in capillary vessel networks

Capillary blood vessels in the body partake in the exchange of gas and nutrients with tissues. They are interconnected via multiple vascular junctions forming the microvascular network. Distributions of blood flow and red cells (RBCs) in such networks are spatially uneven and vary in time. Since they dictate the pathophysiology of tissues, their knowledge is important. Theoretical models used to obtain flow and RBC distribution in large networks have limitations as they treat each vessel as a one-dimensional segment and do not explicitly consider cell-cell and cell-vessel interactions. High-fidelity computational models that accurately model each individual RBC are computationally too expensive to predict haemodynamics in large vascular networks and over a long time. Here we investigate the applicability of machine learning (ML) techniques to predict blood flow and RBC distributions in physiologically realistic vascular networks. We acquire data from high-fidelity simulations of deformable RBC suspension flowing in the networks. With the flow and haematocrit specified at an inlet of vasculature, the ML models predict the time-averaged flow rate and RBC distributions in the entire network, time-dependent flow rate and haematocrit in each vessel and vascular bifurcation in isolation over a long time, and finally, simultaneous spatially and temporally evolving quantities through the vessel hierarchy in the networks.

Erysense, a Lab-on-a-Chip-Based Point-of-Care Device to Evaluate Red Blood Cell Flow Properties With Multiple Clinical Applications

In many medical disciplines, red blood cells are discovered to be biomarkers since they "experience" various conditions in basically all organs of the body. Classical examples are diabetes and hypercholesterolemia. However, recently the red blood cell distribution width (RDW), is often referred to, as an unspecific parameter/marker (e.g., for cardiac events or in oncological studies). The measurement of RDW requires venous blood samples to perform the complete blood cell count (CBC). Here, we introduce Erysense, a lab-on-a-chip-based point-of-care device, to evaluate red blood cell flow properties. The capillary chip technology in combination with algorithms based on artificial neural networks allows the detection of very subtle changes in the red blood cell morphology. This flow-based method closely resembles in vivo conditions and blood sample volumes in the sub-microliter range are sufficient. We provide clinical examples for potential applications of Erysense as a diagnostic tool [here: neuroacanthocytosis syndromes (NAS)] and as cellular quality control for red blood cells [here: hemodiafiltration (HDF) and erythrocyte concentrate (EC) storage]. Due to the wide range of the applicable flow velocities (0.1-10 mm/s) different mechanical properties of the red blood cells can be addressed with Erysense providing the opportunity for differential diagnosis/judgments. Due to these versatile properties, we anticipate the value of Erysense for further diagnostic, prognostic, and theragnostic applications including but not limited to diabetes, iron deficiency, COVID-19, rheumatism, various red blood cell disorders and anemia, as well as inflammation-based diseases including sepsis.

Technologies for measuring red blood cell deformability

Human red blood cells (RBCs) are approximately 8 μm in diameter, but must repeatedly deform through capillaries as small as 2 μm in order to deliver oxygen to all parts of the body. The loss of this capability is associated with the pathology of many diseases, and is therefore a potential biomarker for disease status and treatment efficacy. Measuring RBC deformability is a difficult problem because of the minute forces (∼pN) that must be exerted on these cells, as well as the requirements for throughput and multiplexing. The development of technologies for measuring RBC deformability date back to the 1960s with the development of micropipette aspiration, ektacytometry, and the cell transit analyzer. In the past 10 years, significant progress has been made using microfluidics by leveraging the ability to precisely control fluid flow through microstructures at the size scale of individual RBCs. These technologies have now surpassed traditional methods in terms of sensitivity, throughput, consistency, and ease of use. As a result, these efforts are beginning to move beyond feasibility studies and into applications to enable biomedical discoveries. In this review, we provide an overview of both traditional and microfluidic techniques for measuring RBC deformability. We discuss the capabilities of each technique and compare their sensitivity, throughput, and robustness in measuring bulk and single-cell RBC deformability. Finally, we discuss how these tools could be used to measure changes in RBC deformability in the context of various applications including pathologies caused by malaria and hemoglobinopathies, as well as degradation during storage in blood bags prior to blood transfusions.