Clinical evaluation of a fully electronic microfluidic white blood cell analyzer.

The White Blood Cell (WBC) count is one of the key parameters signaling the health of the immune system. Abnormal WBC counts often signal a systemic insult to the body such as an underlying infection or an adverse side effect to medication. Typically, the blood collected is sent to a central lab for testing, and results come back within hours, which is often inconvenient and may delay time-sensitive diagnosis or treatment. Here, we present the CytoTracker, a fully electronic, microfluidic based instant WBC analyzer with the potential to be used at point-of-care. The CytoTracker is a lightweight, portable, affordable platform capable of quantifying WBCs within minutes using only 50 µl of blood (approximately one drop of blood). In this study, we clinically evaluated the accuracy and performance of CytoTracker in measuring WBC and granulocyte counts. A total of 210 adult patients were recruited in the study. We validated the CytoTracker against a standard benchtop analyzer (Horiba Point of Care Hematology Analyzer, ABX Micros 60). Linear dynamic ranges of 2.5 k/µl- 35 k/µl and 0.6 k/µl- 26 k/µl were achieved for total WBC count and granulocyte count with correlation coefficients of 0.97 and 0.98. In addition, we verified CytoTracker's capability of identifying abnormal blood counts with above 90% sensitivity and specificity. The promising results of this clinical validation study demonstrate the potential for the use of the CytoTracker as a reliable and accurate point-of-care WBC analyzer.

Multi-modal sensing with integrated machine learning to differentiate specific leukocytes targeted by electrically sensitive hybrid particles.

The growing need for personalized, accurate, and non-invasive diagnostic technology has resulted in significant advancements, from pushing current mechanistic limitations to innovative modality developments across various disease-related biomarkers. However, there still lacks clinical solutions for analyzing multiple biomarkers simultaneously, limiting prognosis for patients suffering with complicated diseases or comorbidities. Here, we conceived, fabricated, and validated a multifrequency impedance cytometry apparatus with novel frequency-sensitive barcoded metal oxide Janus particles (MOJPs) as cell-receptor targeting agents. These microparticles are modulated by a metal oxide semi-coating which exhibit electrical property changes in a multifrequency electric field and are functionalized to target CD11b and CD66b membrane proteins on neutrophils. A multi-modal system utilizing supervised machine learning and simultaneous high-speed video microscopy classifies immune-specific surface receptors targeted by MOJPs as they form neutrophil-MOJP conjugates, based on multivariate multifrequency electrical recordings. High precision and sensitivity were determined based on the type of MOJPs conjugated with cells (>90% accuracy between neutrophil-MOJP conjugates versus cells alone). Remarkably, the design could differentiate the number of MOJPs conjugated per cell within the same MOJP class (>80% accuracy); which also improved comparing electrical responses across different MOJP types (>75% accuracy) as well. Such trends were consistent in individual blood samples and comparing consolidated data across multiple samples, demonstrating design robustness. The configuration may further expand to include more MOJP types targeting critical biomarker receptors in one sample and increase the modality's multiplexing potential.

A two-minute assay for electronic quantification of antibodies in saliva enabled through a reusable microfluidic multi-frequency impedance cytometer and machine learning analysis.

The use of saliva as a diagnostic fluid has always been appealing due to the ability for rapid and non-invasive sampling for monitoring health status and the onset and progression of disease and treatment progress. Saliva is rich in protein biomarkers and provides a wealth of information for diagnosis and prognosis of various disease conditions. Portable electronic tools which rapidly monitor protein biomarkers would facilitate point-of-care diagnosis and monitoring of various health conditions. For example, the detection of antibodies in saliva can enable rapid diagnosis and tracking disease pathogenesis of various auto-immune diseases like sepsis. Here, we present a novel method involving immuno-capture of proteins on antibody coated beads and electrical detection of dielectric properties of the beads. The changes in electrical properties of a bead when capturing proteins are extremely complex and difficult to model physically in an accurate manner. The ability to measure impedance of thousands of beads at multiple frequencies, however, allows for a data-driven approach for protein quantification. By moving from a physics driven approach to a data driven approach, we have developed, for the first time ever to the best of our knowledge, an electronic assay using a reusable microfluidic impedance cytometer chip in conjunction with supervised machine learning to quantifying immunoglobulins G (IgG) and immunoglobulins A (IgA) in saliva within two minutes.

Nucleic Acid Quantification by Multi-Frequency Impedance Cytometry and Machine Learning.

Determining nucleic acid concentrations in a sample is an important step prior to proceeding with downstream analysis in molecular diagnostics. Given the need for testing DNA amounts and its purity in many samples, including in samples with very small input DNA, there is utility of novel machine learning approaches for accurate and high-throughput DNA quantification. Here, we demonstrated the ability of a neural network to predict DNA amounts coupled to paramagnetic beads. To this end, a custom-made microfluidic chip is applied to detect DNA molecules bound to beads by measuring the impedance peak response (IPR) at multiple frequencies. We leveraged electrical measurements including the frequency and imaginary and real parts of the peak intensity within a microfluidic channel as the input of deep learning models to predict DNA concentration. Specifically, 10 different deep learning architectures are examined. The results of the proposed regression model indicate that an R_Squared of 97% with a slope of 0.68 is achievable. Consequently, machine learning models can be a suitable, fast, and accurate method to measure nucleic acid concentration in a sample. The results presented in this study demonstrate the ability of the proposed neural network to use the information embedded in raw impedance data to predict the amount of DNA concentration.

Biosensors and machine learning for enhanced detection, stratification, and classification of cells: a review.

Biological cells, by definition, are the basic units which contain the fundamental molecules of life of which all living things are composed. Understanding how they function and differentiating cells from one another, therefore, is of paramount importance for disease diagnostics as well as therapeutics. Sensors focusing on the detection and stratification of cells have gained popularity as technological advancements have allowed for the miniaturization of various components inching us closer to Point-of-Care (POC) solutions with each passing day. Furthermore, Machine Learning has allowed for enhancement in the analytical capabilities of these various biosensing modalities, especially the challenging task of classification of cells into various categories using a data-driven approach rather than physics-driven. In this review, we provide an account of how Machine Learning has been applied explicitly to sensors that detect and classify cells. We also provide a comparison of how different sensing modalities and algorithms affect the classifier accuracy and the dataset size required.

Antibody-functionalized aluminum oxide-coated particles targeting neutrophil receptors in a multifrequency microfluidic impedance cytometer.

Personalized diagnostics of infectious diseases require monitoring disease progression due to their ever-changing physiological conditions and the multi-faceted organ system mechanisms involved in disease pathogenesis. In such instances, the recommended clinical strategies involve multiplexing data collection from critical biomarkers related to a patient's conditions along with longitudinal frequent patient monitoring. Numerous detection technologies exist both in research and commercial settings to monitor these conditions, however, they fail to provide biomarker multiplexing ability with design and data processing simplicity. For a recently conceived multiplexing biomarker modality, this work demonstrates the use of electrically sensitive microparticles targeting and identifying membrane receptors on leukocytes using a single detection source, with a high potential for multiplexing greater than any existing impedance-based single-detection scheme. Here, polystyrene microparticles are coated with varying thicknesses of metal oxides, which generate quantifiable impedance shifts when exposed to multifrequency electric fields depending on the metal oxide thickness. Using multifrequency impedance cytometry, these particles can be measured and differentiated rapidly across one coplanar electrode scheme. After surface-functionalizing particles with antibodies targeting CD11b and CD66b receptors, the particles are combined with isolated neutrophils to measure receptor expression. A combination of data analysis techniques including multivariate analysis, supervised machine learning, and unsupervised machine learning was able to accurately differentiate samples with up to 91% accuracy. This proof-of-concept study demonstrates the potential for these oxide-coated particles for enumerating specific leukocytes enabling multiplexing. Further, additional coating thicknesses or different metal oxide materials can enable a compendium of multiplexing targeting resource to be used to develop a high-multiplexing sensor for targeting membrane receptor expression.

Aluminum Oxide-Coated Particle Differentiation Employing Supervised Machine Learning and Impedance Cytometry.

This article uses a supervised machine learning (ML) system for identifying groups of nanoparticles coated with metal oxides of varying thicknesses using a microfluidic impedance cytometer. These particles generate unique impedance signatures when probed with a multifrequency electric field and finds applications in enabling many multiplexed biosensing technologies. However, current experimental and data processing techniques are unable to sensitively differentiate different metal oxide coated particle types. Here, we employ various machine learning models and collect multiple particle metrics measured. In reported experiments, a 75% accuracy was determined to separate aluminum oxide coated (10nm and 30nm), which is significantly greater than observing only univariate data between different microparticle types. This approach will enable ML models to differentiate such particles with greater accuracies.

MACHINE LEARNING ENABLES QUANTIFYING CELL-JANUS PARTICLE CONJUGATES THROUGH MICROFLOWING IMPEDANCE SIGNALS.

In this work, we demonstrate the differentiation of demodulated multifrequency signals from impedance sensitive microparticles when targeting surface receptors on neutrophils in a microfluidic impedance cytometer. These scheme uses a single signal input and detection configuration, and machine learning can differentiate particle types with up to 82% accuracy.

Multi-frequency impedance sensing for detection and sizing of DNA fragments.

Electronic biosensors for DNA detection typically utilize immobilized oligonucleotide probes on a signal transducer, which outputs an electronic signal when target molecules bind to probes. However, limitation in probe selectivity and variable levels of non-target material in complex biological samples can lead to nonspecific binding and reduced sensitivity. Here we introduce the integration of 2.8 µm paramagnetic beads with DNA fragments. We apply a custom-made microfluidic chip to detect DNA molecules bound to beads by measuring Impedance Peak Response (IPR) at multiple frequencies. Technical and analytical performance was evaluated using beads containing purified Polymerase Chain Reaction (PCR) products of different lengths (157, 300, 613 bp) with DNA concentration ranging from 0.039 amol to 7.8 fmol. Multi-frequency IPR correlated positively with DNA amounts and was used to calculate a DNA quantification score. The minimum DNA amount of a 300 bp fragment coupled on beads that could be robustly detected was 0.0039 fmol (1.54 fg or 4750 copies/bead). Additionally, our approach allowed distinguishing beads with similar molar concentration DNA fragments of different lengths. Using this impedance sensor, purified PCR products could be analyzed within ten minutes to determine DNA fragment length and quantity based on comparison to a known DNA standard.

Functionalization of hybrid surface microparticles for in vitro cellular antigen classification.

Hybrid material surfaces on microparticles are emerging as vehicles for many biomedical multiplexing applications. Functionalization of these hybrid surface microparticles to biomolecules presents unique challenges related to optimization of surface chemistries including uniformity, repeatability, and sample sparring. Hybrid interfaces between microlevel surfaces and individual biomolecules will provide different microenvironments impacting the surface functionalization optimization and efficiency. Here, we propose and validate the first demonstration of streptavidin adsorption-based antibody functionalization on unmodified, hybrid surface microparticles for in vitro analysis. We test this analytical technique and fabricate hybrid surface microparticles with a polystyrene core and aluminum oxide semi-coating. Additionally, we optimize the streptavidin-biotin functionalization chemistry in both assay implementation and sample sparring via analytical mass balances for these microparticles and subsequently conjugate anti-human CD11b antibodies. Result confirmation and characterization occurs from ultraviolet protein absorbance and ImageJ processing of fluorescence microscopy images. Additionally, we design and implement the multi-sectional imaging (MSI) approach to support functionalization uniformity on the hybrid surface microparticles. Finally, as a proof-of-concept performance, we validate anti-CD11b antibodies functionalization by visualizing hybrid surface microparticles conjugate to human neutrophils isolated from blood samples collected from potentially septic patients. Our study introduces and defines a category of functionalization for hybrid surface microparticles with the intent of minuscule sample volumes, low cost, and low environmental impact to be used for many cellular or proteomic in vitro multiplexing applications in the future. Graphical abstract.

Electronic classification of barcoded particles for multiplexed detection using supervised machine learning analysis.

Wearable biosensors are of great interest in recent years due to their potential in health related applications. Multiplex biomarker analysis is needed in wearable devices to improve the sensitivity and reliability. Electronic barcoding of micro-particles has the possibility to enable multiplexed biomarker analysis. Compared with traditional optical and plasmonic methods for barcoding, electronically barcoded particles can be classified using ultra-compact electronic readout platforms. Nano-electronic barcoding works by depositing a thin layer of oxide on the top half of a micro-particle. The thickness and dielectric property of the oxide layer can be tuned to modulate the frequency dependent impedance signature of the particles. A one to one correspondence between a target biomarker and each barcoded particle can potentially be established using this technique. The barcoded particles could be tested with wearable devices to enable multiplex analysis for portable point-of-care diagnostics and real-time monitoring. In this work, we fabricated nine barcoded particles by forming oxide layers of different thicknesses and different dielectric materials using atomic layer deposition and assessed the ability to accurately classify particle barcodes using multi-frequency impedance cytometry in conjunction with supervised machine learning.

Electrical impedance as an indicator of microalgal cell health.

Separating specific cell phenotypes from a heterotypic mixture is a critical step in many research projects. Traditional methods usually require a large sample volume and a complex preparation process that may alter cell property during the sorting process. Here we present the use of electrical impedance as an indicator of cell health and for identifying specific microalgal phenotypes. We developed a microfluidic platform for measuring electrical impedance at different frequencies using the bacterium-sized green alga Picochlorum SE3. The cells were cultured under different salinity conditions and sampled at four different time points. Our results demonstrate the utility of electrical impedance as an indicator of cell phenotype by providing results that are consistent with known changes in cell size and physiology. Outliers in the cell data distribution are particularly useful because they represent phenotypes that have the ability to maintain size and/or membrane ionic permeability under prolonged salt stress. This suggests that our device can be used to identify and sort desired (e.g., experimentally evolved, mutant) cell phenotypes based on their electrical impedance properties.

Toward point-of-care assessment of patient response: a portable tool for rapidly assessing cancer drug efficacy using multifrequency impedance cytometry and supervised machine learning.

We present a novel method to rapidly assess drug efficacy in targeted cancer therapy, where antineoplastic agents are conjugated to antibodies targeting surface markers on tumor cells. We have fabricated and characterized a device capable of rapidly assessing tumor cell sensitivity to drugs using multifrequency impedance spectroscopy in combination with supervised machine learning for enhanced classification accuracy. Currently commercially available devices for the automated analysis of cell viability are based on staining, which fundamentally limits the subsequent characterization of these cells as well as downstream molecular analysis. Our approach requires as little as 20 µL of volume and avoids staining allowing for further downstream molecular analysis. To the best of our knowledge, this manuscript presents the first comprehensive attempt to using high-dimensional data and supervised machine learning, particularly phase change spectra obtained from multi-frequency impedance cytometry as features for the support vector machine classifier, to assess viability of cells without staining or labelling.

Fully integrated wearable impedance cytometry platform on flexible circuit board with online smartphone readout.

We present a wearable microfluidic impedance cytometer implemented on a flexible circuit wristband with on-line smartphone readout for portable biomarker counting and analysis. The platform contains a standard polydimethylsiloxane (PDMS) microfluidic channel integrated on a wristband, and the circuitry on the wristband is composed  of a custom analog lock-in amplification system, a microcontroller with an 8-bit analog-to-digital converter (ADC), and a Bluetooth module wirelessly paired with a smartphone. The lock-in amplification (LIA) system is implemented with a novel architecture which consists of the lock-in amplifier followed by a high-pass filter stage with DC offset subtraction, and a post-subtraction high gain stage enabling detection of particles as small as 2.8  µm using the 8-bit ADC. The Android smartphone application was used to initiate the system and for offline data-plotting and peak counting, and supports online data readout, analysis, and file management. The data is exportable to researchers and medical professionals for in-depth analysis and remote health monitoring. The system, including the microfluidic sensor, microcontroller, and Bluetooth module all fit on the wristband with a footprint of less than 80 cm2. We demonstrate the ability of the system to obtain generalized blood cell counts; however the system can be applied to a wide variety of biomarkers by interchanging the standard microfluidic channel with microfluidic channels designed for biomarker isolation.