Deep learning-assisted ultra-accurate smartphone testing of paper-based colorimetric ELISA assays

A fully automated paper-based smartphone-assisted microfluidic chemiluminescence sample to result immunoassay platform

BACKGROUND

In recent years, adapting immunoassay technologies for point-of-care testing (POCT) to enable low-cost, on-site diagnostics has gained great attention. Among the various approaches to achieving this goal, microfluidic paper-based analytical devices (μPADs) have emerged as a promising platform. These devices are particularly appealing due to their low cost, simple procedures, and minimal sample consumption. Among the detection methods used in μPADs, chemiluminescence (CL) has attracted widespread attention for its high sensitivity. However, existing CL immunoassays often require expensive detection equipment and extensive manual operation, which limits their use in resource-poor areas or where trained personnel are lacking.

RESULTS

In this study, we developed a newly designed μPAD for CL immunoassays, leveraging smartphone-based image capture, valve-regulated reaction process, and app-driven result analysis to enable fully automated detection. Each element of the device was highly integrated and designed for programmable operation, enhancing reliability and stability. An app was developed to manage the entire process, maximizing the convenience of smartphones. The device's effectiveness was calibrated using rabbit IgG as a model, achieving a limit of detection (LOD) of 62.4 pg/mL, which represents an improvement compared to the colorimetric ELISA method with a LOD of 3 ng/mL (20pM). We then applied the device to detect tau protein, a biomarker associated with Alzheimer's disease and achieved a detection limit of 26.1 pg/mL, which is lower than the clinical cut-off value.

SIGNIFICANCE

Our device achieves complete automation and integrates the versatility of μPADs with the precision of CL immunoassays, offering an effective and accessible solution for POCT. By incorporating a user-friendly smartphone interface, it simplifies operation while maintain detection sensitivity, providing a practical and valuable improvement in the development of POCT technologies.

Rapid antimicrobial susceptibility tests performed by self-diluting microfluidic chips for drug resistance studies and point-of-care diagnostics

Antimicrobial resistance (AMR) is a global public health issue. Rapid and accurate antimicrobial susceptibility tests (AST) on bacteria isolates would facilitate appropriate choice of antibiotics, in which patients receive appropriate treatment and the emergence of multidrug-resistant organisms could be prevented simultaneously. In this study, we have developed a microfluidic device named Self Dilution for Faster Antimicrobial Susceptibility Testing (SDFAST). This SlipChip-based device consists of two layers of microchips, allowing injection of bacterial suspension and antibiotics by simply connecting the two chips. By slipping one microchip against another in a single press of the microchip, the antibiotics can be diluted within seconds and be well mixed with bacterial samples. By combining SDFAST with a water-soluble tetrazolium salt-8 (WST-8) assay, a range of clinically prevalent bacteria, including Acinetobacter baumannii, Escherichia coli, Klebsiella pneumoniae, and Staphylococci species, were tested under various antibiotics. Color analysis after 4-6 h of incubation showed an abrupt change in the WST-8 color of certain wells with diluted antibiotics, proving that instrument-free and immediate identification of minimum inhibitory concentration (MIC) could be achieved. The testing on 51 clinical isolates had an agreement of 92%, proving the accuracy of our method. These results validated its advantages of simple operation, rapid testing, and low sample consumption comparing to conventional methods, which require 16-24 h of incubation. Therefore, our method shows great potential to be further developed into a medical instrument for automated medical testing and point-of-care diagnosis.

Mapping human fingerprint beyond level-3 based on an amphiphilic aggregation-induced emission luminogen and the construction of intelligent platform for personal identification

BACKGROUND

Fluorescence imaging agents have been benefiting tremendously from tailor-made aggregation-induced emission (AIE) luminogens, owing to their high on-off ratio, large signal contrast, low background noise as well as the resistance to photobleaching. In the domain of fingerprint imaging, AIE luminogens are beginning to exhibit an advantage owing to the aforementioned superiorities.

RESULTS

We present an amphiphilic benzoic-acid salicylaldehyde AIE luminogen AIE-BASB, and outline its water sensitivity, self-assembly behavior as well as fingerprint imaging properties. AIE-BASB self-assembles into nanoscale textures when fabricated into a drop-casting film but undergoes a disassembly process in response to trace water on fingertip upon physical-contacting. Owing to the biological textures on the skin, fingerprint image can be clearly recorded by AIE-BASB film, which reveals detailed microscopic features of fingerprint information ranging from level-1 to level-3. Furthermore, it allows accurate measurements of the sizes, shapes, centroids, and areas of the sweat pores, which leads the fingerprint information into the next level. In addition, we develop an intelligent system based on AIE-BASB by integrating hardware and software modules, which is capable of recording and identifying fingerprint. After inputting fingerprint segments in trial operation, this intelligent system makes identification by calculation of the categorical probability, and successfully predicts the classification of the undefined fingerprint segments, implying 100 % accuracy in fingerprint identification.

SIGNIFICANCE

We predict that AIE-BASB may facilitate the development of new biometric technologies, which have broad applications in the domain of artificial intelligence, including machine tactility, target perception and object-machine interaction.

TimePAD─Unveiling Temporal Sequence ELISA Signal by Deep Learning for Rapid Readout and Improved Accuracy in a Microfluidic Paper-Based Analytical Platform

The integration of paper-based microfluidics with deep learning represents a pivotal trend in enhancing diagnostic capabilities. This paper introduces a new approach to improve the performance of a paper-based microfluidic enzyme-linked immunosorbent assay (ELISA) by training the temporal sequence colorimetric data rather than static data conventionally, using deep learning. Traditional deep learning-assisted ELISA analysis methods usually rely on a single snapshot of the reaction at its end, which limits the further improvement of sensitivity and specificity (or accuracy for combined evaluation), as it misses dynamic changes in the reaction over time. In this work, we developed a temporal sequence-enhanced paper analytical device (TimePAD) that captures continuous video data of the ELISA reaction, which contains the dynamic colorimetric changes. With the YOLOv8 deep learning alogrithm and the Rabbit IgG as the model for ELISA assay, we can use the initial 20 min signal instead of waiting for 30 min for full reaction, achieving a 33% reduction in the turnaround time. Moreover, the overall accuracy at 20 min is 94.1%, which is slightly improvement to the 93.5% using a traditional single snapshot method at 30 min. This method not only accelerates result interpretation but also enhances the overall efficiency of diagnostics, making it particularly valuable for time-sensitive point-of-care testing applications. Lastly, to demonstrate its real-world use, we expanded to the disease biomarker cTnI detection and obtained accuracy of 98.1% within only 10 min, compared to 25 min with 97.8% accuracy in traditional methods.

Smartphones as a platform for molecular analysis: concepts, methods, devices and future potential

Over the past 15 years, smartphones have had a transformative effect on everyday life. These devices also have the potential to transform molecular analysis over the next 15 years. The cameras of a smartphone, and its many additional onboard features, support optical detection and other aspects of engineering an analytical device. This article reviews the development of smartphones as platforms for portable chemical and biological analysis. It is equal parts conceptual overview, technical tutorial, critical summary of the state of the art, and outlook on how to advance smartphones as a tool for analysis. It further discusses the motivations for adopting smartphones as a portable platform, summarizes their enabling features and relevant optical detection methods, then highlights complementary technologies and materials such as 3D printing, microfluidics, optoelectronics, microelectronics, and nanoparticles. The broad scope of research and key advances from the past 7 years are reviewed as a prelude to a perspective on the challenges and opportunities for translating smartphone-based lab-on-a-chip devices from prototypes to authentic applications in health, food and water safety, environmental monitoring, and beyond. The convergence of smartphones with smart assays and smart apps powered by machine learning and artificial intelligence holds immense promise for realizing a future for molecular analysis that is powerful, versatile, democratized, and no longer just the stuff of science fiction.

An integrated paper-based microfluidic platform for screening of early-stage Alzheimer's disease by detecting Aβ42

Alzheimer's disease (AD) is the leading cause of dementia worldwide, and the development of early screening methods can address its significant health and social consequences. In this paper, we present a rotary-valve assisted paper-based immunoassay device (RAPID) for early screening of AD, featuring a highly integrated on-chip rotary micro-valve that enables fully automated and efficient detection of the AD biomarker (amyloid beta 42, Aβ42) in artificial plasma. The microfluidic paper-based analytical device (μPAD) of the RAPID pre-stores the required assay reagents on a μPAD and automatically controls the liquid flow through a single valve. Once the test sample is added, the test reagents are sequentially transferred to the test area in the order set by the enzyme-linked immunosorbent assay (ELISA) protocol. In addition, the RAPID can remotely control the operation of the μPAD valve via a micro-servomotor, quantify the signals generated, display the results, and wirelessly transmit the data to a smartphone. To calibrate the RAPID, we performed a sandwich ELISA for Aβ42 in artificial plasma, and obtained a low limit of detection (LOD) of 9.6 pg mL-1, a coefficient of determination (COD) of 0.994, and an individual assay time of ∼30 minutes. In addition, we simulated 24 artificial samples to quantify Aβ42 protein concentrations in artificial plasma samples. The results show good consistency between the conventional ELISA and RAPID detection. The experimental results demonstrate that the RAPID is expected to promote further popularization of the screening of early-stage AD.

Multi-Task Water Quality Colorimetric Detection Method Based on Deep Learning

The colorimetric method, due to its rapid and low-cost characteristics, demonstrates a wide range of application prospects in on-site water quality testing. Current research on colorimetric detection using deep learning algorithms predominantly focuses on single-target classification. To address this limitation, we propose a multi-task water quality colorimetric detection method based on YOLOv8n, leveraging deep learning techniques to achieve a fully automated process of "image input and result output". Initially, we constructed a dataset that encompasses colorimetric sensor data under varying lighting conditions to enhance model generalization. Subsequently, to effectively improve detection accuracy while reducing model parameters and computational load, we implemented several improvements to the deep learning algorithm, including the MGFF (Multi-Scale Grouped Feature Fusion) module, the LSKA-SPPF (Large Separable Kernel Attention-Spatial Pyramid Pooling-Fast) module, and the GNDCDH (Group Norm Detail Convolution Detection Head). Experimental results demonstrate that the optimized deep learning algorithm excels in precision (96.4%), recall (96.2%), and mAP50 (98.3), significantly outperforming other mainstream models. Furthermore, compared to YOLOv8n, the parameter count and computational load were reduced by 25.8% and 25.6%, respectively. Additionally, precision improved by 2.8%, recall increased by 3.5%, mAP50 enhanced by 2%, and mAP95 rose by 1.9%. These results affirm the substantial potential of our proposed method for rapid on-site water quality detection, offering new technological insights for future water quality monitoring.

Smartphone based lateral flow immunoassay quantifications

Lateral Flow Immunoassay (LFI) is a disposable tool designed to detect target substances using minimal resources. For qualitative analysis, LFI does not require a device (i.e., reader) to interpret test results. However, various studies have been conducted to implement quantitative analysis using LFI systems, incorporating LFI along with electrical/electronic readers, to overcome the limitations associated with qualitative LFI analysis. The reader used for the quantitative analysis of LFI should ensure mobility for easy on-site diagnostics and inspections, be user-friendly in operation, and have a fast processing speed until the results are obtained. Due to these requirements, smartphones are increasingly utilized as readers in quantitative analysis of LFI. Among the various components constituting a smartphone, high-performance cameras can serve as sensors converting visual signals into electrical signals. With powerful processing units, large storage capacity, and network capabilities for transmitting analysis results, smartphones are also utilized as interfaces for quantitative analysis. Absolutely, the widespread global use of smartphones is a key advantage, leading to their utilization as diagnostic devices for acquiring, analyzing, storing, and transmitting assay test results. This paper summarizes research cases where smartphones are utilized as readers for quantitative LFI systems used in confirming contamination in food or the environment, detecting drugs, and diagnosing diseases in humans or animals. The systems are classified based on the types of label particles used in the assay, and efforts to improve the quantitative analysis performance for each are examined. Cases where smartphones were used as LFI readers for the diagnosis of the 2019 Coronavirus Disease (COVID-19), which has recently caused significant global damage, have also been investigated.

Smartphone-based colorimetric detection platform using color correction algorithms to reduce external interference

Smartphone-based digital image colorimetry is a powerful, fast, low-cost approach to detecting target analytes. However, lighting conditions and camera parameters easily affect the detection results, significantly curtailing its applicability in multiple scenarios. In this study, an Android-based mobile application (SMP-CC) is developed, which offers a comprehensive package that includes image acquisition, color correction, and colorimetric analysis functions. Using a custom color card, a built-in algorithm in SMP-CC can minimize the color difference between the standard color block image captured by different smartphones under different lighting conditions and the standard value by an LS171 colorimeter less than 4.36. The algorithm significantly eliminates the impacts of external lighting conditions and differences in cell phone models. Furthermore, the feasibility of SMP-CC was verified by successful colorimetric detection of urine pH, glucose, and protein, demonstrating its potential in smartphone-based digital image colorimetry.

Recent Advances in Microfluidics for the Early Detection of Plant Diseases in Vegetables, Fruits, and Grains Caused by Bacteria, Fungi, and Viruses

In the context of global population growth expected in the future, enhancing the agri-food yield is crucial. Plant diseases significantly impact crop production and food security. Modern microfluidics offers a compact and convenient approach for detecting these defects. Although this field is still in its infancy and few comprehensive reviews have explored this topic, practical research has great potential. This paper reviews the principles, materials, and applications of microfluidic technology for detecting plant diseases caused by various pathogens. Its performance in realizing the separation, enrichment, and detection of different pathogens is discussed in depth to shed light on its prospects. With its versatile design, microfluidics has been developed for rapid, sensitive, and low-cost monitoring of plant diseases. Incorporating modules for separation, preconcentration, amplification, and detection enables the early detection of trace amounts of pathogens, enhancing crop security. Coupling with imaging systems, smart and digital devices are increasingly being reported as advanced solutions.

AI-Assisted Real-Time Immunoassay Improves Clinical Sensitivity and Specificity

Real-time biosensing systems can interrogate the association between the analyte and the biorecognition element across time. Typically, the resulting data are preprocessed to offer valuable bioanalytical information obtained at a single optimal point of such a real-time response; for instance, a diagnosis of certain medical conditions can be established depending on a biomarker (analyte) concentration measured at an optimal time, that is, a threshold. Exploiting this conventional approach, we previously developed a nanophotonic immunoassay for bacterial vaginosis diagnosis exhibiting a clinical sensitivity and specificity of ca. 96.29% (n = 162). Herein, we demonstrate that a real-time biosensing platform assisted by artificial intelligence not only obviates biomarker concentration (i.e., a threshold) determination but also increases sensitivity and specificity in the targeted diagnostic, thereby reaching values of up to 100%.

Automated Offline Smartphone-Assisted Microfluidic Paper-Based Analytical Device for Biomarker Detection of Alzheimer's Disease

This paper presents a smartphone-assisted microfluidic paper-based analytical device (μPAD), which was applied to detect Alzheimer's disease biomarkers, especially in resource-limited regions. This device implements deep learning (DL)-assisted offline smartphone detection, eliminating the requirement for large computing devices and cloud computing power. In addition, a smartphone-controlled rotary valve enables a fully automated colorimetric enzyme-linked immunosorbent assay (c-ELISA) on μPADs. It reduces detection errors caused by human operation and further increases the accuracy of μPAD c-ELISA. We realized a sandwich c-ELISA targeting β-amyloid peptide 1-42 (Aβ 1-42) in artificial plasma, and our device provided a detection limit of 15.07 pg/mL. We collected 750 images for the training of the DL YOLOv5 model. The training accuracy is 88.5%, which is 11.83% higher than the traditional curve-fitting result analysis method. Utilizing the YOLOv5 model with the NCNN framework facilitated offline detection directly on the smartphone. Furthermore, we developed a smartphone application to operate the experimental process, realizing user-friendly rapid sample detection.

Automatic offline-capable smartphone paper-based microfluidic device for efficient biomarker detection of Alzheimer's disease

BACKGROUND

Alzheimer's disease (AD) is a prevalent neurodegenerative disease with no effective treatment. Efficient and rapid detection plays a crucial role in mitigating and managing AD progression. Deep learning-assisted smartphone-based microfluidic paper analysis devices (μPADs) offer the advantages of low cost, good sensitivity, and rapid detection, providing a strategic pathway to address large-scale disease screening in resource-limited areas. However, existing smartphone-based detection platforms usually rely on large devices or cloud servers for data transfer and processing. Additionally, the implementation of automated colorimetric enzyme-linked immunoassay (c-ELISA) on μPADs can further facilitate the realization of smartphone μPADs platforms for efficient disease detection.

RESULTS

This paper introduces a new deep learning-assisted offline smartphone platform for early AD screening, offering rapid disease detection in low-resource areas. The proposed platform features a simple mechanical rotating structure controlled by a smartphone, enabling fully automated c-ELISA on μPADs. Our platform successfully applied sandwich c-ELISA for detecting the β-amyloid peptide 1-42 (Aβ 1-42, a crucial AD biomarker) and demonstrated its efficacy in 38 artificial plasma samples (healthy: 19, unhealthy: 19, N = 6). Moreover, we employed the YOLOv5 deep learning model and achieved an impressive 97 % accuracy on a dataset of 1824 images, which is 10.16 % higher than the traditional method of curve-fitting results. The trained YOLOv5 model was seamlessly integrated into the smartphone using the NCNN (Tencent's Neural Network Inference Framework), enabling deep learning-assisted offline detection. A user-friendly smartphone application was developed to control the entire process, realizing a streamlined "samples in, answers out" approach.

SIGNIFICANCE

This deep learning-assisted, low-cost, user-friendly, highly stable, and rapid-response automated offline smartphone-based detection platform represents a good advancement in point-of-care testing (POCT). Moreover, our platform provides a feasible approach for efficient AD detection by examining the level of Aβ 1-42, particularly in areas with low resources and limited communication infrastructure.

Rapid Microfluidic Immuno-Biosensor Detection System for the Point-of-Care Determination of High-Sensitivity Urinary C-Reactive Protein

A microfluidic immuno-biosensor detection system consisting of a microfluidic spectrum chip and a micro-spectrometer detection device is presented for the rapid point-of-care (POC) detection and quantification of high-sensitivity C-reactive protein (hs-CRP) in urine. The detection process utilizes a highly specific enzyme-linked immunosorbent assay (ELISA) method, in which capture antibodies and detection antibodies are pre-deposited on the substrate of the microchip and used to form an immune complex with the target antigen. Horseradish peroxidase (HRP) is added as a marker enzyme, followed by a colorimetric reaction using 3,3',5,5'-tetramethylbenzidine (TMB). The absorbance values (a.u.) of the colorimetric reaction compounds are measured using a micro-spectrometer device and used to measure the corresponding hs-CRP concentration according to the pre-established calibration curve. It is shown that the hs-CRP concentration can be determined within 50 min. In addition, the system achieves recovery rates of 93.8-106.2% in blind water samples and 94.5-104.6% in artificial urine. The results showed that the CRP detection results of 41 urine samples from patients with chronic kidney disease (CKD) were highly consistent with the conventional homogeneous particle-enhanced turbidimetric immunoassay (PETIA) method's detection results (R2 = 0.9910). The experimental results showed its applicability in the detection of CRP in both urine and serum. Overall, the results indicate that the current microfluidic ELISA detection system provides an accurate and reliable method for monitoring the hs-CRP concentration in point-of-care applications.

Machine learning-assisted image label-free smartphone platform for rapid segmentation and robust multi-urinalysis

This study presents a groundbreaking approach for the early detection of chronic kidney disease (CKD) and other urological disorders through an image-label-free, multi-dipstick identification method, eliminating the need for complex machinery, label libraries, or preset coordinates. Our research successfully identified reaction pads on 187 multi-dipsticks, each with 11 pads, leveraging machine learning algorithms trained on human urine data. This technique aims to surpass traditional colourimetric methods and concentration-colour curve fitting, offering more robust and precise community screening and home monitoring capabilities. The developed algorithms enhance the generalizability of machine learning models by extracting primary colours and correcting urine colours on each reaction pad. This method's cost-effectiveness and portability are significant, as it requires no additional equipment beyond a standard smartphone. The system's performance rivals professional medical equipment without auxiliary lighting or flash under regular indoor light conditions, effectively managing false positives and negatives across various categories with remarkable accuracy. In a controlled experimental setting, we found that random forest algorithms, based on a Bagging strategy and applied in the HSV colour space, showed optimal results in smartphone-assisted urinalysis. This study also introduces a novel urine colour correction method, significantly improving machine learning model performance. Additionally, ISO parameters were identified as crucial factors influencing the accuracy of smartphone-based urinalysis in the absence of additional lighting or optical configurations, highlighting the potential of this technology in low-resource settings.

On-site colorimetric food spoilage monitoring with smartphone embedded machine learning

Real-time and on-site food spoilage monitoring is still a challenging issue to prevent food poisoning. At the onset of food spoilage, microbial and enzymatic activities lead to the formation of volatile amines. Monitoring of these amines with conventional methods requires sophisticated, costly, labor-intensive, and time consuming analysis. Here, anthocyanins rich red cabbage extract (ARCE) based colorimetric sensing system was developed with the incorporation of embedded machine learning in a smartphone application for real-time food spoilage monitoring. FG-UV-CD100 films were first fabricated by crosslinking ARCE-doped fish gelatin (FG) with carbon dots (CDs) under UV light. The color change of FG-UV-CD100 films with varying ammonia vapor concentrations was captured in different light sources with smartphones of various brands, and a comprehensive dataset was created to train machine learning (ML) classifiers to be robust and adaptable to ambient conditions, resulting in 98.8% classification accuracy. Meanwhile, the ML classifier was embedded into our Android application, SmartFood++, enabling analysis in about 0.1 s without internet access, unlike its counterpart using cloud operation via internet. The proposed system was also tested on a real fish sample with 99.6% accuracy, demonstrating that it has a great advantage as a potent tool for on-site real-time monitoring of food spoilage by non-specialized personnel.

Review of paper-based microfluidic analytical devices for in-field testing of pathogens

Pathogens cause various infectious diseases and high morbidity and mortality which is a global public health threat. The highly sensitive and specific detection is of significant importance for the effective treatment and intervention to minimise the impact. However, conventional detection methods including culture and molecular method gravely depend on expensive equipment and well-trained skilled personnel, limiting in the laboratory. It remains challenging to adapt in resource-limiting areas, e.g., low and middle-income countries (LMICs). To this end, low-cost, rapid, and sensitive detection tools with the capability of field testing e.g., a portable device for identification and quantification of pathogens, has attracted increasing attentions. Recently, paper-based microfluidic analytical devices (μPADs) have shown a promising tool for rapid and on-site diagnosis, providing a cost-effective and sensitive analytical approach for pathogens detection. The fast turn-round data collection may also contribute to better understanding of the risks and insights on mitigation method. In this paper, critical developments of μPADs for in-field detection of pathogens both for clinical diagnostics and environmental surveillance are reviewed. The future development, and challenges of μPADs for rapid and onsite detection of pathogens are discussed, including using the cross-disciplinary development with, emerging techniques such as deep learning and Internet of Things (IoT).

Deep Learning for Microfluidic-Assisted Caenorhabditis elegans Multi-Parameter Identification Using YOLOv7

The Caenorhabditis elegans (C. elegans) is an ideal model organism for studying human diseases and genetics due to its transparency and suitability for optical imaging. However, manually sorting a large population of C. elegans for experiments is tedious and inefficient. The microfluidic-assisted C. elegans sorting chip is considered a promising platform to address this issue due to its automation and ease of operation. Nevertheless, automated C. elegans sorting with multiple parameters requires efficient identification technology due to the different research demands for worm phenotypes. To improve the efficiency and accuracy of multi-parameter sorting, we developed a deep learning model using You Only Look Once (YOLO)v7 to detect and recognize C. elegans automatically. We used a dataset of 3931 annotated worms in microfluidic chips from various studies. Our model showed higher precision in automated C. elegans identification than YOLOv5 and Faster R-CNN, achieving a mean average precision (mAP) at a 0.5 intersection over a union (mAP@0.5) threshold of 99.56%. Additionally, our model demonstrated good generalization ability, achieving an mAP@0.5 of 94.21% on an external validation set. Our model can efficiently and accurately identify and calculate multiple phenotypes of worms, including size, movement speed, and fluorescence. The multi-parameter identification model can improve sorting efficiency and potentially promote the development of automated and integrated microfluidic platforms.