Category: ☆☆☆☆☆

Time series anomaly detection is challenging due to the complexity and variety of patterns that can occur. One major difficulty arises from modeling time-dependent relationships to find contextual anomalies while maintaining detection accuracy for point anomalies. In this paper, we propose a framework for unsupervised time series anomaly detection that utilizes point-based and sequence-based reconstruction models. The point-based model attempts to quantify point anomalies, and the sequence-based model attempts to quantify both point and contextual anomalies. Under the formulation that the observed time point is a two-stage deviated value from a nominal time point, we introduce a nominality score calculated from the ratio of a combined value of the reconstruction errors. We derive an induced anomaly score by further integrating the nominality score and anomaly score, then theoretically prove the superiority of the induced anomaly score over the original anomaly score under certain conditions. Extensive studies conducted on several public datasets show that the proposed framework outperforms most state-of-the-art baselines for time series anomaly detection.

We coin our method Nominality score conditioned time series anomaly detection by Point/Sequential Reconstruction (NPSR).

Related Publications

1. C.-Y. Lai, F.-K. Sun, Z. Gao, J. H. Lang, D. S. Boning, Nominality Score Conditioned Time Series Anomaly Detection by Point/Sequential Reconstruction, Neural Information Processing Systems (NeurIPS), 2023.
   [arXiv] [NeurIPS] [GitHub]
2. C.-Y. Lai, F.-K. Sun, J. H. Lang, D. S. Boning, Unsupervised Multivariate Time Series Anomaly Detection for High-Frequency Data, Microsystems Annual Research Conference (MARC), (2023).
   [proceedings]

This study serves as a major extension for my previous research Impedimetric Microfluidic Chip for MUC1 Aptasensing. Similar to the previous one, microfluidic chips are fabricated for impedimetric detection of tumor marker MUC1. Moreover, I integrated an interdigitated array electrode (IDA electrode) into the chip, and realized real-time detection for the aptasensor. In-depth investigation of the relationship between electrochemical properties and microfluidic conditions are also carried out. This system possesses several advantages, such as the highly sensitive characteristic of IDA electrodes, real-time detection, low sample usage, label-free detection using EIS, and miniaturized volume using a microfluidic chip.

Introduction

MUC1 is a surface glycoprotein which over-expresses in several types of cancer cells, making it an ideal marker for cancer detection. For MUC1 recognition, the S2.2 aptamer is a 25mer ssDNA selected in vitro and can affinitively and specifically bind to certain motifs within the MUC1 protein. Electrochemical impedance spectroscopy (EIS) has been proven as an effective method for ultrasensitive MUC1 aptasensing and cell detection [1]. Despite the high sensitivity and selectivity of EIS and the S2.2 aptamer, long reaction times and large sample volumes have hindered EIS biosensors for realistic bioanalysis.

The integration of microfluidics with EIS has a large potential for lowering the amount of usage during reactions and to meet real-time, portable, large-scale and high-throughput requirements. Though up to date, there hasn’t yet been studies regarding real-time impedimetric aptasensing to the best of our knowledge. Thus, this research is dedicated to develop a real-time microfluidic impedance aptasensing platform for affinitive and selective detection of MUC1.

Related Publications

1. C.-Y. Lai, J.-H. Weng, L.-C. Chen, Real-time impedimetric MUC1 aptasensor using microfluidic symmetric Au electrodes, The Twenty Second International Conference on Miniaturized Systems for Chemistry and Life Sciences (µTAS), (2018).
   [poster pdf]
2. J.-H. Weng, C.-Y. Lai, L.-C. Chen, Microfluidic amperometry with two symmetric Au microelectrodes under one-way and shuttle flow conditions, Electrochimica Acta, 297 (2019) 118-128.
   https://doi.org/10.1016/j.electacta.2018.11.128

This is the research topic for my master’s thesis, which serves as an integrated work for the related researches and projects that I have worked on. Relevant studies have been accepted as 4 journal papers (1^st author x2, co-author x2) and 5 international conference papers (1^st author x4, co-author x1). The list below shows all the related topics covered in my thesis:

(Ch. 3) Diffusion Impedance Modeling of Interdigitated Array Electrodes
(Ch. 3.4.6) Electrochemical Impedance Circuit Fitting Program
(Ch. 4) Impedimetric Aptasensing using a Symmetric Randles Circuit Model
(Ch. S.7) Real-time Impedimetric MUC1 Aptasensor using Microfluidic Symmetric Gold Electrode
(Ch. S.8) Real-time Impedance Detection Systems

Figure 1. Research framework for this thesis.

Abstract

The inhibition of tumor markers has been a popular research object among the academic society. They are often detected using simple and low-cost techniques such as electrochemical impedance spectroscopy (EIS), which aptamers are occasionally used as the sensing element for achieving high sensitivity and selectivity. This integrated method has flourished in recent years.

However, for electrochemical methods, a three electrode setup faces fabrication complexity, high cost and low yield rates during miniaturization. Two electrode impedimetric detection using interdigitated array (IDA) electrodes also faces a problem. Due to its geometry, there hasn’t been any studies that derive its diffusion impedance according to different bandwidths and gap widths. Therefore, this study makes a basis on impedimetric modeling of symmetric two electrode systems.

The first part focuses on the derivation and verification of an integral form of solution for IDA diffusion impedance. An equivalent circuit fitting program succeeded to accurately fit the EIS data and parameters such as the ratio of electrode bandwidth to gap width and diffusion coefficient can also be obtained by fitting the data from a single EIS experiment. This can aid researchers in relevant fields model their systems more accurately.

In the second part, a symmetric equivalent circuit model is developed, and it is applied it for impedimetric detection of thrombin and a tumor marker MUC1 with a fabricated aptasensor using standard Au electrodes (SGE) and IDA chips. The model is proved of correctness, and is applied for bio-detection. IDA chips are used for aptasensor fabrication for thrombin detection. The program designed in the first part is used for circuit fitting of EIS data, and accurate parameters are obtained. This sensor has the regenerability for six times of detection and the specificity is also confirmed.

Symmetric Au electrode systems have simple and low fabrication cost characteristics. Its integration with highly stable aptamers can contribute to mass production and customization in product commercialization. According to the above results, the author anticipates future developments in relevant medical diagnosis and point-of-care applications.

Related Publications

1. (Upcoming conference with paper accepted) C.-Y. Lai, T.-H. He, W.-C. Huang, L.-C. Chen, MUC1 impedimetric aptasensing based on interdigitated array electrode chip using a novel diffusion element, 30th Anniversary World Congress on Biosensors, (2020).
2. Master’s thesis: Electrochemical Impedance Modeling of Symmetric Electrodes and Interdigitated Array Chips for Aptasensing Applications

While I was working with interdigitated array electrodes (IDA electrodes) for my previous research Real-time Impedimetric MUC1 Aptasensor using Microfluidic Symmetric Gold Electrode, I discovered a major problem with its impedance spectrum when viewing the Nyquist plot. Until then, there hadn’t been any studies that can explain the diffusion phenomenon of redox species between the band electrodes due to an applied sinusoidal voltage wave, and no circuit elements appear to be suitable for modeling the diffusion impedance of IDA electrodes. Therefore, I devised a theory for modeling an IDA electrodes impedimetric properties using conformal mapping and cylindrical finite length approximation methods, and succeeded to construct a circuit element that can replace the Warburg element, and extract important parameters from the corresponding equivalent circuit model. This can aid researchers in relevant fields to model their systems more accurately.

Here’s a video for visualizing the diffusion phenomenon between the band electrodes of an IDA electrode. COMSOL is used for simulating the time-dependent concentration field of redox species.

Abstract

An analytical problem for impedimetric sensing is usually encountered when using interdigitated array (IDA) electrodes. Finite diffusion of redox species dominates at low frequencies and confuses researchers, making incorrect understanding of underlying phenomena possible. In this work, an integral equation for calculating the diffusion impedance of IDA electrodes is derived using conformal mapping and cylindrical finite length approximation. Electrodes of different bandwidths and gap widths are fabricated, and their heights and symmetric electrochemical characteristics are verified. Simulations are performed to verify the predicted constant concentration contours. The calculated zero-frequency impedance showed high correlation with the reciprocal of limiting current calculated from literature study (R² = 0.992) and from chronoamperometry experiments (R² = 0.970). Further evidence for the correctness of theory is established due to the fact that experimental EIS data and calculated impedances are highly consistent (R² ≥ 0.948 for real and imaginary part). This sheds some light on explaining the diffusion phenomenon of impedance using IDA electrodes in the low frequency spectrum. An equivalent circuit fitting program is further designed for fitting several elements including the IDA electrode diffusion impedance derived in the theory. The program succeeded to accurately fit the EIS data (average MSE = 0.611), which using the Warburg element failed (average MSE = 54.86). Parameters such as the ratio of electrode bandwidth to gap width and diffusion coefficient can also be determined by fitting the data from a single EIS experiment. Another impedance calculation program is also given, which can aid researchers in relevant fields to model their systems more accurately.

Related Publications

1. C.-Y. Lai, J.-H. Weng, W.-L. Shih, L.-C. Chen, C.-F. Chou, P.-K. Wei, Diffusion impedance modeling for interdigitated array electrodes by conformal mapping and cylindrical finite length approximation, Electrochimica Acta, 320 (2019) 134629.
   https://doi.org/10.1016/j.electacta.2019.134629
2. C.-Y. Lai, J.-H. Weng, W.-L. Shih, L.-C. Chen, C.-F. Chou, P.-K. Wei, Diffusion impedance modeling for interdigitated array electrodes by conformal mapping and cylindrical finite length approximation, 11th International Symposia on Electrochemical Impedance Spectroscopy, (2019).
   [abstract] [presentation pdf] [presentation clip]

This is the research topic for my bachelor’s thesis. After completing the Surface Plasmon Resonance Platform project, I acquired microfabrication techniques for fabricating microfluidic gold surface chips. The gold surface can be patterned to form a two-electrode setup, where electrochemical detection can be applied for biosensing. For this project, I fabricated a microfluidic microelectrode chip that can detect the concentration of the tumor marker MUC1 using a DNA aptamer. A method called electrochemical impedance spectroscopy (EIS) is used for detecting the change of impedance caused by different concentrations of MUC1. I won the Outstanding Poster Award for poster presentation of this work at the 64^th TwIChE (Taiwan Institute of Chemical Engineers) Annual Meeting.

Introduction

Mucin 1(MUC1) is a glycoprotein which plays an important role within the human body such as cell signaling and defense of pathogens. Overexpression of MUC1 on cancer cell surfaces promotes cell survival and tumor angiogenesis. Therefore, the detection of MUC1 has become a new trend in early detection of cancer and therapy. Aptamers are synthesized DNA or RNA which serves as suitable sensing elements for targets such as MUC1 or other proteins and can achieve excellent affinity and selectivity. Electrochemical impedance spectroscopy (EIS) is a powerful electrochemical detection technique regarding its ability for monitoring different stages during the fabrication process of aptasensors. Moreover, the label-free, simple and highly sensitive features further strengthen its potential of integrating with other applications. Microfluidics are recently recognized as a technique advantageous for performing bioanalysis. Not only can they lower the amount of usage during reactions, but also have the potential to meet real-time, portable, large-scale and high-throughput requirements. In this study, a novel aptasensor for detecting MUC1 using microfluidic integrated gold electrode is fabricated and demonstrated. Parameter fitting of the equivalent circuit is performed after each step during the fabrication and protein detection. An LOD of 0.42nM is achieved and the fitting result suggests several varying element parameters influenced by the concentration of MUC1. The simple, label-free and low reagent usage detection of MUC1 using this microfluidic impedimetric aptasensor is achieved.

Related Publications

1. C.-Y. Lai, L.-C. Chen, The study of an impedimetric microfluidic chip design for mucin1 aptasensing, 64th TwIChE Annual Meeting, (2017). Outstanding Poster Award
   [poster pdf]
2. C.-Y. Lai, L.-C. Chen, Detection of mucin1 with a microfluidic impedimetric aptasensor, International Symposium on Smart-Sensing Medical Devices and 22nd Symposium of Association for Chemical Sensors in Taiwan, (2017).
   [abstract] [presentation pdf] [presentation clip]
3. Bachelor’s thesis: The Study of an Impedimetric Microfluidic Chip Applied for MUC1 Aptasensing