Image-based feedback and analysis system for digital microfluidics

Dynamic video recognition for cell-encapsulating microfluidic droplets

Droplet microfluidics is a highly sensitive and high-throughput technology extensively utilized in biomedical applications, such as single-cell sequencing and cell screening. However, its performance is highly influenced by the droplet size and single-cell encapsulation rate (following random distribution), thereby creating an urgent need for quality control. Machine learning has the potential to revolutionize droplet microfluidics, but it requires tedious pixel-level annotation for network training. This paper investigates the application software of the weakly supervised cell-counting network (WSCApp) for video recognition of microdroplets. We demonstrated its real-time performance in video processing of microfluidic droplets and further identified the locations of droplets and encapsulated cells. We verified our methods on droplets encapsulating six types of cells/beads, which were collected from various microfluidic structures. Quantitative experimental results showed that our approach can not only accurately distinguish droplet encapsulations (micro-F1 score > 0.94), but also locate each cell without any supervised location information. Furthermore, fine-tuning transfer learning on the pre-trained model also significantly reduced (>80%) annotation. This software provides a user-friendly and assistive annotation platform for the quantitative assessment of cell-encapsulating microfluidic droplets.

Open Hardware for Microfluidics: Exploiting Raspberry Pi Singleboard Computer and Camera Systems for Customisable Laboratory Instrumentation

The integration of Raspberry Pi miniature computer systems with microfluidics has revolutionised the development of low-cost and customizable analytical systems in life science laboratories. This review explores the applications of Raspberry Pi in microfluidics, with a focus on imaging, including microscopy and automated image capture. By leveraging the low cost, flexibility and accessibility of Raspberry Pi components, high-resolution imaging and analysis have been achieved in direct mammalian and bacterial cellular imaging and a plethora of image-based biochemical and molecular assays, from immunoassays, through microbial growth, to nucleic acid methods such as real-time-qPCR. The control of image capture permitted by Raspberry Pi hardware can also be combined with onboard image analysis. Open-source hardware offers an opportunity to develop complex laboratory instrumentation systems at a fraction of the cost of commercial equipment and, importantly, offers an opportunity for complete customisation to meet the users' needs. However, these benefits come with a trade-off: challenges remain for those wishing to incorporate open-source hardware equipment in their own work, including requirements for construction and operator skill, the need for good documentation and the availability of rapid prototyping such as 3D printing plus other components. These advances in open-source hardware have the potential to improve the efficiency, accessibility, and cost-effectiveness of microfluidic-based experiments and applications.

Enhancing Microdroplet Image Analysis with Deep Learning

Microfluidics is a highly interdisciplinary field where the integration of deep-learning models has the potential to streamline processes and increase precision and reliability. This study investigates the use of deep-learning methods for the accurate detection and measurement of droplet diameters and the image restoration of low-resolution images. This study demonstrates that the Segment Anything Model (SAM) provides superior detection and reduced droplet diameter error measurement compared to the Circular Hough Transform, which is widely implemented and used in microfluidic imaging. SAM droplet detections prove to be more robust to image quality and microfluidic images with low contrast between the fluid phases. In addition, this work proves that a deep-learning super-resolution network MSRN-BAM can be trained on a dataset comprising of droplets in a flow-focusing microchannel to super-resolve images for scales ×2, ×4, ×6, ×8. Super-resolved images obtain comparable detection and segmentation results to those obtained using high-resolution images. Finally, the potential of deep learning in other computer vision tasks, such as denoising for microfluidic imaging, is shown. The results show that a DnCNN model can denoise effectively microfluidic images with additive Gaussian noise up to σ = 4. This study highlights the potential of employing deep-learning methods for the analysis of microfluidic images.

WSCNet: Biomedical Image Recognition for Cell Encapsulated Microfluidic Droplets

Microfluidic droplets accommodating a single cell as independent microreactors are frequently demanded for single-cell analysis of phenotype and genotype. However, challenges exist in identifying and reducing the covalence probability (following Poisson's distribution) of more than two cells encapsulated in one droplet. It is of great significance to monitor and control the quantity of encapsulated content inside each droplet. We demonstrated a microfluidic system embedded with a weakly supervised cell counting network (WSCNet) to generate microfluidic droplets, evaluate their quality, and further recognize the locations of encapsulated cells. Here, we systematically verified our approach using encapsulated droplets from three different microfluidic structures. Quantitative experimental results showed that our approach can not only distinguish droplet encapsulations (F1 score > 0.88) but also locate each cell without any supervised location information (accuracy > 89%). The probability of a "single cell in one droplet" encapsulation is systematically verified under different parameters, which shows good agreement with the distribution of the passive method (Residual Sum of Squares, RSS < 0.5). This study offers a comprehensive platform for the quantitative assessment of encapsulated microfluidic droplets.

Image-based real-time feedback control of magnetic digital microfluidics by artificial intelligence-empowered rapid object detector for automated in vitro diagnostics

In vitro diagnostics (IVD) plays a critical role in healthcare and public health management. Magnetic digital microfluidics (MDM) perform IVD assays by manipulating droplets on an open substrate with magnetic particles. Automated IVD based on MDM could reduce the risk of accidental exposure to contagious pathogens among healthcare workers. However, it remains challenging to create a fully automated IVD platform based on the MDM technology because of a lack of effective feedback control system to ensure the successful execution of various droplet operations required for IVD. In this work, an artificial intelligence (AI)-empowered MDM platform with image-based real-time feedback control is presented. The AI is trained to recognize droplets and magnetic particles, measure their size, and determine their location and relationship in real time; it shows the ability to rectify failed droplet operations based on the feedback information, a function that is unattainable by conventional MDM platforms, thereby ensuring that the entire IVD process is not interrupted due to the failure of liquid handling. We demonstrate fundamental droplet operations, which include droplet transport, particle extraction, droplet merging and droplet mixing, on the MDM platform and show how the AI rectify failed droplet operations by acting upon the feedback information. Protein quantification and antibiotic resistance detection are performed on this AI-empowered MDM platform, and the results obtained agree well with the benchmarks. We envision that this AI-based feedback approach will be widely adopted not only by MDM but also by other types of digital microfluidic platforms to offer precise and error-free droplet operations for a wide range of automated IVD applications.

Digital microfluidic platform assembled into a home-made studio for sample preparation and colorimetric sensing of S-nitrosocysteine

Digital microfluidics (DMF) is a versatile lab-on-a-chip platform that allows integration with several types of sensors and detection techniques, including colorimetric sensors. Here, we propose, for the first time, the integration of DMF chips into a mini studio containing a 3D-printed holder with previously fixed UV-LEDs to promote sample degradation on the chip surface before a complete analytical procedure involving reagent mixture, colorimetric reaction, and detection through a webcam integrated on the equipment. As a proof-of-concept, the feasibility of the integrated system was successfully through the indirect analysis of S-nitrosocysteine (CySNO) in biological samples. For this purpose, UV-LEDs were explored to perform the photolytic cleavage of CySNO, thus generating nitrite and subproducts directly on DMF chip. Nitrite was then colorimetrically detected based on a modified Griess reaction, in which reagents were prepared through a programable movement of droplets on DMF devices. The assembling and the experimental parameters were optimized, and the proposed integration exhibited a satisfactory correlation with the results acquired using a desktop scanner. Under the optimal experimental conditions, the obtained CySNO degradation to nitrite was 96%. Considering the analytical parameters, the proposed approach revealed linear behavior in the CySNO concentration range between 12.5 and 400 μmol L^-1 and a limit of detection equal to 2.8 μmol L^-1. Synthetic serum and human plasma samples were successfully analyzed, and the achieved results did not statistically differ from the data recorded by spectrophotometry at the confidence level of 95%, thus indicating the huge potential of the integration between DMF and mini studio to promote complete analysis of lowmolecular weight compounds.

Integrating machine learning and digital microfluidics for screening experimental conditions

Digital microfluidics (DMF) has the signatures of an ideal liquid handling platform - as shown through almost two decades of automated biological and chemical assays. However, in the current state of DMF, we are still limited by the number of parallel biological or chemical assays that can be performed on DMF. Here, we report a new approach that leverages design-of-experiment and numerical methodologies to accelerate experimental optimization on DMF. The integration of the one-factor-at-a-time (OFAT) experimental technique with machine learning algorithms provides a set of recommended optimal conditions without the need to perform a large set of experiments. We applied our approach towards optimizing the radiochemistry synthesis yield given the large number of variables that affect the yield. We believe that this work is the first to combine such techniques which can be readily applied to any other assays that contain many parameters and levels on DMF.

Vision-Based Performance Analysis of an Active Microfluidic Droplet Generation System Using Droplet Images

This paper discusses an active droplet generation system, and the presented droplet generator successfully performs droplet generation using two fluid phases: continuous phase fluid and dispersed phase fluid. The performance of an active droplet generation system is analysed based on the droplet morphology using vision sensing and digital image processing. The proposed system in the study includes a droplet generator, camera module with image pre-processing and identification algorithm, and controller and control algorithm with a workstation computer. The overall system is able to control, sense, and analyse the generation of droplets. The main controller consists of a microcontroller, motor controller, voltage regulator, and power supply. Among the morphological features of droplets, the diameter is extracted from the images to observe the system performance. The MATLAB-based image processing algorithm consists of image acquisition, image enhancement, droplet identification, feature extraction, and analysis. RGB band filtering, thresholding, and opening are used in image pre-processing. After the image enhancement, droplet identification is performed by tracing the boundary of the droplets. The average droplet diameter varied from ~3.05 mm to ~4.04 mm in the experiments, and the average droplet diameter decrement presented a relationship of a second-order polynomial with the droplet generation time.

Microfluidic Devices Controlled by Machine Learning with Failure Experiments

A critical microchannel technique is to isolate specific objects, such as cells, in a biological solution. Generally, this particle sorting is achieved by designing a microfluidic device and tuning its control values; however, unpredictable motions of the particle mixture make this approach time-consuming and labor intensive. Here, we show that microfluidic control with reinforced learning characterized by utilizing failure results can maximize the training effect with limited data. This method uses microscopic images of the separation process, including failed conditions (inappropriate flow speeds or dilution rates of biological samples), and insights for efficient learning are provided by setting gradient rewards according to the degree of failure. Once learning is performed in this manner, the optimal separating condition for other related samples can be automatically found. Failed experiments are not wasteful; they increase training data and make it easier to reach correct answers. This device control could be useful in automatic synthetic chemistry, biomedical analysis, and microfabrication robotics.

Bacterial classification and antibiotic susceptibility testing on an integrated microfluidic platform

With the prevalence of bacterial infections and increasing levels of antibiotic resistance comes the need for rapid and accurate methods for bacterial classification (BC) and antibiotic susceptibility testing (AST). Here we demonstrate the use of the fluid handling technique digital microfluidics (DMF) for automated and simultaneous BC and AST using growth metabolic markers. Custom instrumentation was developed for this application including an integrated heating module and a machine-learning-enabled low-cost colour camera for real-time absorbance and fluorescent sample monitoring on multipurpose devices. Antibiotic dilutions along with sample handling, mixing and incubation at 37 °C were all pre-programmed and processed automatically. By monitoring the metabolism of resazurin, resorufin beta-D-glucuronide and resorufin beta-D-galactopyranoside to resorufin, BC and AST were achieved in under 18 h. AST was validated in two uropathogenic E. coli strains with antibiotics ciprofloxacin and nitrofurantoin. BC was performed independently and simultaneously with ciprofloxacin AST for E. coli, K. pneumoniae, P. mirabilis and S. aureus. Finally, a proof-of-concept multiplexed system for breakpoint testing of two antibiotics, as well as E. coli and coliform classification was investigated with a multidrug-resistant E. coli strain. All bacteria were correctly identified, while AST and breakpoint test results were in essential and category agreement with reference methods. These results show the versatility and accuracy of this all-in-one microfluidic system for analysis of bacterial growth and phenotype.

Expanding the limits towards 'one-pot' DNA assembly and transformation on a rapid-prototype microfluidic device

DNA assembly and transformation are crucial to the building process in synthetic biology. These steps are significant roadblocks when engineering increasingly complex biological systems. To address this, recent development of widespread 'biofoundry' facilities has employed automation equipment to expedite the synthetic biology workflow. Despite significant progress, there is a clear demand for lower-cost and smaller-footprint automation equipment. The field of microfluidics have emerged to provide automation capabilities to meet this demand. However, we still lack devices capable of building large multi-gene systems in a consolidated process. In response to this challenge, we have developed a digital microfluidic platform that performs "one-pot" Golden Gate DNA assembly of large plasmids and transformation of E coli. The system features a novel electrode geometry and modular design, which make these devices simple to fabricate and use, thus improving the accessibility of microfluidics. This device incorporates an impedance-based adaptive closed loop water replenishment system to compensate for droplet evaporation and maintain constant assembly reaction concentrations, which we found to be crucial to the DNA assembly efficiency. We also showcase a closed-loop temperature control system that generates precise thermodynamic profiles to optimize heat shock transformation. Moreover, we validated the system by assembling and transforming large and complex plasmids conferring a biosynthetic pathway, resulting in performance comparable to those of standard techniques. We propose that the methods described here will contribute to a new generation of accessible automation platforms aimed at speeding up the 'building' process, lowering reagent consumption and removing manual work from synthetic biology.

Digital Microfluidics Chips for the Execution and Real-Time Monitoring of Multiple Ribozymatic Cleavage Reactions

In this paper, we describe the design and performance of two digital microfluidics (DMF) chips capable of executing multiple ribozymatic reactions, with proper controls, in response to short single-stranded DNA inducers. Since the fluorescence output of a reaction is measurable directly from the chip, without the need for gel electrophoresis, a complete experiment involving up to eight reactions (per chip) can be carried out reliably, relatively quickly, and efficiently. The ribozymes can also be used as biosensors of the concentration of oligonucleotide inputs, with high sensitivity, low limits of quantification and of detection, and excellent signal-to-noise ratio. The presented chips are readily usable devices that can be used to automate, speed up, and reduce the costs of ribozymatic reaction experiments.

Machine vision-based driving and feedback scheme for digital microfluidics system

A digital microfluidic system based on electrowetting-on-dielectric is a new technology for controlling microliter-sized droplets on a plane. By applying a voltage signal to an electrode, the droplets can be controlled to move, merge, and split. Due to device design, fabrication, and runtime uncertainties, feedback control schemes are necessary to ensure the reliability and accuracy of a digital microfluidic system for practical application. The premise of feedback is to obtain accurate droplet position information. Therefore, there is a strong need to develop a digital microfluidics system integrated with driving, position, and feedback functions for different areas of study. In this article, we propose a driving and feedback scheme based on machine vision for the digital microfluidics system. A series of experiments including droplet motion, merging, status detection, and self-adaption are performed to evaluate the feasibility and the reliability of the proposed scheme. The experimental results show that the proposed scheme can accurately locate multiple droplets and improve the success rate of different applications. Furthermore, the proposed scheme provides an experimental platform for scientists who focused on the digital microfluidics system.

High-resolution Arduino-based data acquisition devices for microscale separation systems

In this work, we have designed, constructed, and evaluated simple, inexpensive open-source data acquisition systems based on various analog-to-digital converter modules (ADS 1115, MCP 3424, LTC 2400, with resolution from 16 to 24-bit) and a miniature Arduino Nano ™ microcontroller. The constructed data acquisition systems provide excellent performance and are comparable to a commercial, 24-bit device. We provide full schematics and corresponding source codes so that analytical chemists can easily construct any of the developed systems without extensive electronic or programming knowledge. The 24-bit LTC 2400 based device provided the best and comparable performance to a commercial, high-end 24-bit sigma to delta converter (ORCA 2800) at a fraction of cost (less than 50 USD compared to 870 USD for the commercial counterpart). The excellent performance was verified using a capillary electrophoresis system with contactless conductivity detection and separation of inorganic ions in clinical skin wipe and tap water samples.

Feedback-controlled microbubble generator producing one million monodisperse bubbles per second

Monodisperse lipid-coated microbubbles are a promising route to unlock the full potential of ultrasound contrast agents for medical diagnosis and therapy. Here, we present a stand-alone lab-on-a-chip instrument that allows microbubbles to be formed with high monodispersity at high production rates. Key to maintaining a long-term stable, controlled, and safe operation of the microfluidic device with full control over the output size distribution is an optical transmission-based measurement technique that provides real-time information on the production rate and bubble size. We feed the data into a feedback loop and demonstrate that this system can control the on-chip bubble radius (2.5 μm-20 μm) and the production rate up to 10⁶ bubbles/s. The freshly formed phospholipid-coated bubbles stabilize after their formation to a size approximately two times smaller than their initial on-chip bubble size without loss of monodispersity. The feedback control technique allows for full control over the size distribution of the agent and can aid the development of microfluidic platforms operated by non-specialist end users.

One Cell, One Drop, One Click: Hybrid Microfluidics for Mammalian Single Cell Isolation

Generating a stable knockout cell line is a complex process that can take several months to complete. In this work, a microfluidic method that is capable of isolating single cells in droplets, selecting successful edited clones, and expansion of these isoclones is introduced. Using a hybrid microfluidics method, droplets in channels can be individually addressed using a co-planar electrode system. In the hybrid microfluidics device, it is shown that single cells can be trapped and subsequently encapsulate them on demand into pL-sized droplets. Furthermore, droplets containing single cells are either released, kept in the traps, or merged with other droplets by the application of an electric potential to the electrodes that is actuated through an in-house user interface. This high precision control is used to successfully sort and recover single isoclones to establish monoclonal cell lines, which is demonstrated with a heterozygous NCI-H1299 lung squamous cell population resulting from loss-of-function eGFP and RAF1 gene knockout transfections.

Electrowetting-on-Dielectric System for COVID-19 Testing

The ongoing viral outbreak labeled COVID-19 is spreading rapidly across states and is posing a great threat to public health. Rapid identification of the virus in the population plays a crucial role in isolating the individual and breaking the transmission chain, apart from initiating an appropriate treatment procedure. Here, we discuss an electrowetting-on-dielectric (EWOD) technology that uses a microprocessor-controlled electrode array to merge a possibly infected sample carried by a liquid drop with a drop of a reagent to carry out the testing process. Changes in color occurring during the mixing process of the drops are imaged using a camera.

A fucosyltransferase inhibition assay using image-analysis and digital microfluidics

Sialyl-LewisX and LewisX are cell-surface glycans that influence cell-cell adhesion behaviors. These glycans are assembled by α(1,3)-fucosyltransferase enzymes. Their increased expression plays a role in inflammatory disease, viral and microbial infections, and cancer. Efficient screens for specific glycan modifications such as those catalyzed by fucosyltransferases are tended toward costly materials and large instrumentation. We demonstrate for the first time a fucosylation inhibition assay on a digital microfluidic system with the integration of image-based techniques. Specifically, we report a novel lab-on-a-chip approach to perform a fluorescence-based inhibition assay for the fucosylation of a labeled synthetic disaccharide, 4-methylumbelliferyl β-N-acetyllactosaminide. As a proof-of-concept, guanosine 5'-diphosphate has been used to inhibit Helicobacter pylori α(1,3)-fucosyltransferase. An electrode shape (termed "skewed wave") is designed to minimize electrode density and improve droplet movement compared to conventional square-based electrodes. The device is used to generate a 10 000-fold serial dilution of the inhibitor and to perform fucosylation reactions in aqueous droplets surrounded by an oil shell. Using an image-based method of calculating dilutions, referred to as "pixel count," inhibition curves along with IC₅₀ values are obtained on-device. We propose the combination of integrating image analysis and digital microfluidics is suitable for automating a wide range of enzymatic assays.

Micro-Electrode-Dot-Array Digital Microfluidic Biochips: Technology, Design Automation, and Test Techniques

Digital microfluidic biochips (DMFBs) are being increasingly used for DNA sequencing, point-of-care clinical diagnostics, and immunoassays. DMFBs based on a micro-electrode-dot-array (MEDA) architecture have recently been proposed, and fundamental droplet manipulations, e.g., droplet mixing and splitting, have also been experimentally demonstrated on MEDA biochips. There can be thousands of microelectrodes on a single MEDA biochip, and the fine-grained control of nanoliter volumes of biochemical samples and reagents is also enabled by this technology. MEDA biochips offer the benefits of real-time sensitivity, lower cost, easy system integration with CMOS modules, and full automation. This review paper first describes recent design tools for high-level synthesis and optimization of map bioassay protocols on a MEDA biochip. It then presents recent advances in scheduling of fluidic operations, placement of fluidic modules, droplet-size-aware routing, adaptive error recovery, sample preparation, and various testing techniques. With the help of these tools, biochip users can concentrate on the development of nanoscale bioassays, leaving details of chip optimization and implementation to software tools.

A low-cost, open-source digital stripchart recorder for chromatographic detectors using a Raspberry Pi

One of the most critical aspects of chromatographic analysis is effective data acquisition and processing. Typical approaches include software platforms designed for specific instruments or commercial data acquisition hardware boards, both of which require expensive licenses to use and operate. To increase the access and affordability of chromatographic data acquisition, especially for systems in which software control has become obsolete or must be written in-house, an open-source digital stripchart recorder has been developed. This system is built upon a Raspberry Pi single-board computer and a plug-in printed circuit board with the necessary integrated circuits for data acquisition. Using an open-source software called Processing, a complete user interface to control the system was developed that enables the acquisition, filtering, and processing of chromatographic data. The system performs comparably to more expensive platforms, with calculated values for peak area, retention time, and plate count all within 3% of the values calculated by a widely used commercial chromatography data software package.

An integrated droplet-digital microfluidic system for on-demand droplet creation, mixing, incubation, and sorting

Droplet microfluidics is a technique that has the ability to compartmentalize reactions in sub nano- (or pico-) liter volumes that can potentially enable millions of distinct biological assays to be performed on individual cells. In a typical droplet microfluidic system, droplets are manipulated by pressure-based flows. This has limited the fluidic operations that can be performed in these devices. Digital microfluidics is an alternative microfluidic paradigm with precise control and manipulation over individual droplets. Here, we implement an integrated droplet-digital microfluidic (which we call 'ID2M') system in which common fluidic operations (i.e. droplet generation, cell encapsulation, droplet merging and mixing, droplet trapping and incubation, and droplet sorting) can be performed. With the addition of electrodes, we have been able to create droplets on-demand, tune their volumes on-demand, and merge and mix several droplets to produce a dilution series. Moreover, this device can trap and incubate droplets for 24 h that can consequently be sorted and analyzed in multiple n-ary channels (as opposed to typical binary channels). The ID2M platform has been validated as a robust on-demand screening system by sorting fluorescein droplets of different concentration with an efficiency of ∼96%. The utility of the new system is further demonstrated by culturing and sorting tolerant yeast mutants and wild-type yeast cells in ionic liquid based on their growth profiles. This new platform for both droplet and digital microfluidics has the potential to be used for screening different conditions on-chip and for applications like directed evolution.

An Automated Induction Microfluidics System for Synthetic Biology

The expression of a recombinant gene in a host organism through induction can be an extensively manual and labor-intensive procedure. Several methods have been developed to simplify the protocol, but none has fully replaced the traditional IPTG-based induction. To simplify this process, we describe the development of an autoinduction platform based on digital microfluidics. This system consists of a 600 nm LED and a light sensor to enable the real-time monitoring of  the optical density (OD) samples coordinated with the semicontinuous mixing of a bacterial culture. A hand-held device was designed as a microbioreactor to culture cells and to measure the OD of the bacterial culture. In addition, it serves as a platform for the analysis of regulated protein expression in E. coli without the requirement of standardized well-plates or pipetting-based platforms. Here, we report for the first time, a system that offers great convenience without the user to physically monitor the culture or to manually add inducer at specific times. We characterized our system by looking at several parameters (electrode designs, gap height, and growth rates) required for an autoinducible system. As a first step, we carried out an automated induction optimization assay using a RFP reporter gene to identify conditions suitable for our system. Next, we used our system to identify active thermophilic β-glucosidase enzymes that may be suitable candidates for biomass hydrolysis. Overall, we believe that this platform may be useful for synthetic biology applications that require regulating and analyzing expression of heterologous genes for strain optimization.