Towards an Automated MEMS-based Characterization of Benign and Cancerous Breast Tissue using Bioimpedance Measurements

Abstract

Micro-Electro-Mechanical-Systems (MEMS) are desirable for use within medical diagnostics because of their capacity to manipulate and analyze biological materials at the microscale. Biosensors can be incorporated into portable lab-on-a-chip devices to quickly and reliably perform diagnostics procedure on laboratory and clinical samples. In this paper, electrical impedance-based measurements were used to distinguish between benign and cancerous breast tissues using microchips in a real-time and label-free manner. Two different microchips having inter-digited electrodes (10 µm width with 10 µm spacing and 10 µm width with 30 µm spacing) were used for measuring the impedance of breast tissues. The system employs Agilent E4980A precision impedance analyzer. The impedance magnitude and phase were collected over a frequency range of 100 Hz to 2 MHz. The benign group and cancer group showed clearly distinguishable impedance properties. At 200 kHz, the difference in impedance of benign and cancerous breast tissue was significantly higher (3110 Ω) in the case of microchips having 10 µm spacing compared to microchip having 30 µm spacing (568 Ω).

Keywords: MEMS sensor, Inter-digited Electrodes, Bioimpedance, Breast Cancer, Tissue Micro Array

Go to:

1. Introduction

In recent years, breast cancer is the most common type of cancer in women and one of the leading causes of female mortality worldwide [1] and the majority of patients having breast cancer diagnosis have the sub-type invasive ductal carcinoma [2]. There is a need to develop newer and better diagnostic tools for the detection of breast cancer and bioimpedance is hypothesized as one such approach in this work. Bioimpedance is defined as the opposition of a biological tissue to current flow. Bioimpedance analysis can be broadly defined as the impedance of biological specimens that includes a wide spectrum of applications ranging from the whole human body measurements to those conducted at the level of DNA. Since the middle of the 20th century, electrical impedance measurement technique has become a standard analytic tool used in the early-recognition of disease onset, 3D monitoring of the cancer progression, detection of pathogenic bacteria in food, and dynamic identification and quantification of cellular changes [3–6]. The utility of this technology originates from the fact that impedance can capture and reflect the rich biological and electrical characteristics exhibited at the whole body, organ, tissue, cell and organelle level [7]. The recent development of micro/nanotechnology has provided additional advantages and introduces new paradigms into the investigation of traditional electrical impedance concepts. The biosensors fabricated using MEMS technology can be used for real-time monitoring of the underlying mechanisms of the onset and progression of cancer [8]. With the advent of micro and nanofabrication technologies, conventional bench-top instruments can be miniaturized [8]. Similarly, microfluidic embedded sensor systems for achieving integrated systems, such as micro total analysis system (µTAS) is also possible [9]. Miniaturized sensors have significant advantages such as high throughput, small characteristic dimensions, low power consumption, and portability [10]. Drug screening [11], cellular adhesion [12], and cellular kinetics [13] have all been successfully monitored using bioimpedance techniques. Inter-digitated electrodes (IDEs) are implemented in various sensing devices such as surface acoustic wave (SAW) sensors, chemical sensors as well as MEMS based biosensors [14]. MEMS-based bioimpedance measurement are primarily limited to cells [11,12,14]. Different techniques such as microfilters [15–18], microwells [19–21] and dam structure [22] have been implemented for separation of cells. Automated characterization of cancerous cells using micro-cavity array and micromanipulator for exact positioning of cellular spheroids has been studied by other groups [11,23,24]. Ozkan et al. [25] and Umehara et al. [26] showed optical tweezers for manipulation of cells in micro-fluidic devices, including parallel manipulation of cells. Microdevices with modified interior surface by reactive coating and microwells using SU-8 were studied to dock the cells within predefined location [19, 27–29]. Quartz crystal microbalance (QCM) sensor has been used to study the cell-surface interactions and different cell-adhesion behaviors between the healthy and ill-behaved cells [5, 30–31]. Electrode Cell substrate Impedance Sensing (ECIS) is a non-invasive method used to the study the cells based on their attachment to the substrate (electrodes) [30]. This method has been recently used to study cell properties such as attachment, spreading, growth and proliferation [30, 32–38]. While IDEs have been optimized for a variety of single cell-based sensing applications, the impedance characterization at the tissue level, in particular, benign and cancerous breast tissue has not been adequately investigated. The advantage of characterizing micro-scale tissues over cells is that the degree of malignancy as well as the architectural changes that occur during the progression of cancer can be quantified on a larger micro-scale sample to make a deterministic assessment compared to studying a single cell.

An automated measurement system for bio-medical detection has advantages such as higher efficiency, excellent repeatability and statistical reliability [39]. Yamamoto et al. developed an automated system for measuring the real part (electrical resistive component) and imaginary part (reactive component) of the impedance [40]. Our goal is to build an automated bioimpedance measurement system using MEMS-based devices capable of measuring bioimpedance of tissue array simultaneously thereby increasing the throughput. To achieve this we have made an array of microdevices (IDE’s) [5 × 6] glass wafer. The semiautomated way of impedance measurement is as shown in Fig. 1. In this work, we have fabricated a microchip having interdigital electrodes inside a SU-8 well to measure the impedance of benign and cancerous breast tissues. To our knowledge, this is the first such study for characterizing the benign and cancerous breast tissues using IDEs and bioimpedance method. It is known that the output signal strength of microchips having IDEs can be controlled through careful design of the active area, width, and spacing of the electrode fingers [41, 42]. In this paper, we investigated breast tissue specimen that originated from a total of ten cases of high-grade invasive ductal carcinoma. The “benign breast tissues” are sampled from normal appearing breast lobules or terminal ducts, and “cancerous breast tissues” are high grade invasive ductal carcinoma. The rest of this paper is organized as follows. In section 2, we provide the experimental details for sensor fabrication, tissue microarray preparation and lumped circuit modeling of the microchip-tissue interface. In section 3, we provide the experimental results and discussion. Finally, in section 4, we make some concluding remarks. The schematics of (a) the MEMS-based bioimpedance sensor for tissue microarray and (b) the semi-automated bioimpedance measurement system is shown in Fig.1.

Open in a separate window

Fig. 1

Schematics of (a) the MEMS-based bio-impedance sensor for tissue microarray and (b) the automated bio-impedance measurement system.

Go to:

2. Experimental Work

2.1 Sensor Fabrication

To make the sensor biocompatible, the sensor was fabricated on 1.0 mm-thick glass substrate (Pyrex 7740). The microchip was fabricated using a two-mask process. The fabrication process is shown in Fig. 2. After cleaning glass using a 70:30 solution of H2SO4 and H2O2 [Fig. 2(a)], Cr/Au (0.1 µm/0.5 µm) was deposited using thermal evaporation on glass substrates [Fig. 2(b)]. Using a standard photolithography technique interdigital electrodes (10 µm width with 10 µm spacing and 30 µm spacing) were patterned [Fig. 2(c)]. SU-8 2007 was used to form reservoir. SU-8 photoresist was spin coated at 2000 rpm for 1 minute on glass substrate to obtain 7 µm thickness [Fig. 2(d)] and was patterned using photolithography to form reservoir [Fig. 2(e)]. For using SU-8 on glass substrates, adhesion promoters are not required. The reservoir is made to hold the tissue and medium inside the desired area. The schematic of the microchip is shown in Fig. 3(a) and the reservoir dimensions are shown in Fig. 3(b). Figure 4(a) and 4(b) shows the optical photographs of the fabricated microchips with 10 µm and 30 µm spacing, respectively. The width in both cases is kept at 10 µm.

Open in a separate window

Fig. 2

Process flow for the microchip.

Fig. 3

Schematic of the microchip.

Fig. 4

Optical photographs of (a) microchip with IDEs having 10 µm width and 10 µm spacing and (b) 10 µm width and 30 µm spacing.

2.2 Tissue Microarray Preparation

Benign and cancerous breast tissue blocks were carefully selected from archival tissue resources at the Histopathology and Imaging Core Facility at Rutgers Cancer Institute of New Jersey. Conventional histology specimen preparation protocol was used in this experiment to fix, slice, and deparaffinize the tissue.

Briefly, fresh tissue is first fixed in formalin – this process preserves tissue from degradation and maintains cellular structure and sub-cellular structure. The fixed specimen is then dehydrated by going through a series of progressively more concentrated ethanol, followed by hydrophobic clearing agent (xylene), infiltration agent (molten paraffin wax), and finally embedded in liquid wax. After cooling to room temperature, the paraffin-embedded specimen can be stored for a very long period of time (a tissue bank in this case), and is hard enough to be sliced at 4 µm for histological purposes. The deparaffinization protocol is also routine histological process, in which the wax is first softened in a baking oven and then gradually replaced with water through the series of ethanol baths. The microchips were deparaffinized as follows: Xylene (5 min, 3 times); 100% Alcohol (5 min, 2 times); 95% Alcohol (5 min, 1 time); 75% Alcohol (5 min, 1 time); rinse in PBS (1–2 times) and immersed in PBS holding solution until impedance analysis. Following the fixation and preparation process described above, the Histology and Imaging Core Facility at Rutgers Cancer Institute of New Jersey archives them and hosts a cancer related tissue bank by storing the formalin-fixed paraffin-embedded (FFPE) specimens at room temperature. Our archive, along with similar tissue banks at other institutes, supported countless experiments in explorative cancer research.

The conventional FFPE method to preserve tissue had been proven effective in reproducing cell morphology and tissue architecture after decades of storage. It is also considered effective in preserving molecular antigenicity for immunohistochemical studies in majority of the cases. Protocols also exist to extract DNA and RNA sample from FFPE specimens for investigative use.

Uniform-sized tissue cores of 0.6 mm in diameter were extracted from each donor tissue block, and inserted into recipient paraffin blocks using Manual Tissue Arrayer (Beecher Instruments) [43]. These single-cored TMAs were rigorously quality-controlled via Hematoxylin & Eosin (H&E) stained sections and the above process was repeated until two characteristic tissue cores were identified from each case. The individual tissue cores were sectioned at 4 µm in thickness and carefully positioned in the center of the interdigited electrodes. Fig. 5 shows the optical photograph of cancerous tissue placed on the interdigited electrodes having 30 µm spacing.

Fig. 5

Optical photograph of cancerous tissue placed on interdigited electrodes: (a) 4X and (b) 20X magnification.

As any geometrical variations of tissue samples can affect the impedance measurement, the following quality assurance steps and checks were implemented. The microtome (Reichert-Jung 2030) used in the experiment is regularly maintained by professionals to ensure accuracy. Additional slices of tissue were fixed onto coverglass and examined by SEM imaging. Figure 6 shows the size (diameter) of the single tissue core as being 0.6 mm. A cross-sectional SEM image of the tissue shown in Fig. 7 shows that the thickness of the tissue is 4.0 ± 0.17 µm. All specimens in one experimental batch were cut and transferred to devices in one session and hence the sample preparation condition was constant in terms of microtome thickness setting and water bath temperature. During the course of preparing and conducting the experiment, the tissue cores were quality assured multiple times under the microscope to ensure completeness of the circular shape and flat attachment to the device.

Fig. 6

SEM image of breast tissue cores.

Fig. 7

Cross-sectional image of the breast tissue core.

2.3 Circuit Modeling

The microchip interface is modeled using a lumped circuit model as shown in Fig. 8. The equivalent circuit consists of the PBS solution resistance RPBS, the double layer capacitance at the electrode-solution interface Cdl and the tissue impedance Zt in a series connection [44]. The tissue impedance Zt is modeled as an ohmic resistance Rt and a capacitance Ct element in parallel [45]. The net series impedance is given by:

Znet = Xdt + RPBS‖Zt + Xdt

(1)

where,

Z

t

=

Rt

1+j2πf

Rt

C

t

and

X

dt

=

1

j2πf

C

dt

Open in a separate window

Fig. 8

Schematic of (a) bio-impedance measurement of tissues and (b) equivalent electrical circuit.

Where, f is the frequency, Xdl is the capacitive reactance due to the electrical double layer and 

j=

−1−−−

√

Substituting values of Zt and Xdl in Eq. 1 we get,

Z

net

=

2

j2πf

C

dl

+(

R

PBS

∥∥∥Rt

1+j2πf

Rt

C

t

)

(2)

Simplifying Eq. 2, we get:

Z

net

=

RtR

PBS

(

Rt

+

R

PBS

)

(

Rt

+

R

PBS

)

2

4

π

2

f

2R2t

C

2tR2

PBS

+1

−j[

1

πf

C

dt

+

2πf

C

tR2tR2

PBS

(

Rt

+

R

PBS

)

2

4

π

2

f

2R2t

C

2tR2

PBS

+1

]

(3)

The circuit parameters are estimated by fitting Eq. 3 to the experimental data. Optimal parameter estimates should describe both impedance and phase data; hence residuals from both impedance and phase need to be accounted in the fitting approach. We solve the following non-linear regression problem to estimate the circuit parameters:

Minimize=log[

∑ni=1

(

Z

expt

−|

Z

net

|)

2

]

+log[(

∑ni=1

(

ϕexpi

−arg|

Z

net

|)

2

)]

(4)

Where, 

Z

expt

 and ϕexpt

 are the measured impedance and phase respectively and n is number of data points. The Nelson-Mead simplex algorithm [46] is used to solve the regression problem described by Eq. 4. The logarithm of the sum of squared residuals in Eq. 4 serves the purpose of scaling down the residuals in impedance and phase. This leads to numerical stability in fitting the experimental data to Eq. 3, and reduces the dependence on accurate choice of the starting point of the simplex fitting algorithm.

Go to:

3. Results and Discussion

Breast lesions include many different classes of pathologies, among which invasive ductal carcinoma is the most common form of breast cancer. In this work, we investigated differences between high grade invasive ductal carcinoma with tumor adjacent areas and benign morphology. Impedance measurements were performed for the uniform-sized benign and cancerous breast tissue cores placed on IDEs using an Agilent E4980A impedance analyzer. The impedance analyzer was calibrated and operated using the following steps: (a) Connect the probes and turn the power ON, (b) press “Measurement Setup”, (c) press “Correction” and test for open measurement and short measurement, (d) once the open and short measurements are done, press “Measurement Setup” (e) press function key and select “More” option. Select: z → z-θd, frequency → 2 MHz, Level DC = 10 mV, Trigger → manual, measurement time → Medium, average = 16, (f) press “Display Format”. Select “List Sweep” and press “Trigger”.

A voltage of 10 mV was applied to the contact pads using gold probe tips. No DC bias was applied. The impedance magnitude and phase were collected over a frequency range of 100 Hz to 2 MHz.

The impedance and the phase shift was measured for benign and cancerous breast tissue on microchip having IDE with 10 µm spacing and 30 µm spacing and are shown in Fig. 9 and Fig. 10 respectively. It was observed that the impedance and phase shift for both benign as well as cancerous breast tissue placed on IDEs with 30 µm spacing were higher than those of tissue placed on IDEs with 10 µm spacing. Furthermore, the differences between benign and cancerous tissues were more prominent when examined with the 10 µm spacing devices than on 30 µm spacing devices (3110 Ω vs 568.0 Ω at 200 KHz respectively).

Fig. 9

Plots of impedance measurement for (a) benign breast tissue and (b) cancerous breast tissue placed on microchips with IDEs having 10 µm and 30 µm spacing, respectively.

Fig. 10

Plots of phase measurement for (a) benign breast tissue and (b) cancerous breast tissue placed on microchips with IDEs having 10 µm and 30 µm spacing, respectively.

The slope covering the majority of the frequency spectrum is characteristic of the double layer capacitance at the electrode-solution interface Cdl and the tissue impedance Zt in a series connection [32]. As the impedance is inversely proportional to the capacitance, the impedance decreases when the capacitance increases on increasing the surface area. It is consistent with our observation that the impedance increases when the spacing between the IDE fingers increase.

Impedance measurements were performed on a total of eight specimens, which included 4 benign specimens and 4 cancerous specimens. Two representative cores were extracted from each specimen. Microchip having IDEs with 10 µm spacing was used for a total of 16 measurements. As seen from Fig. 11, benign and cancerous breast tissues have clear differences in their impedance and phase, which can be identified using microchip and impedance analyzer. Note that the benign breast tissues (8023-1, 8023-2) and cancerous breast tissues (3343-7, 3343-8) were acquired from older archived tissues (over 8 years in storage), which may have affected their bioimpedance characteristics.

Fig. 11

Plots of (a) impedance and (b) phase measurement for benign and cancerous breast tissues using microchip having IDEs with 10 µm spacing.

Prior to our first attempt to study impedance properties of FFPE specimen, we were not sure what affect the preservation method or the length of storage had in change of impedance and we had to hypothesize that it had minimal or limited effect. It can be observed from Fig. 11 that: (i) the duplicated core from each case shows similar characteristics indicating the stability of measurement, (ii) with increase in frequency, the impedance of the tissue decreases and (iii) impedance properties aggregate well within the benign and cancerous group while the distinction are clear between groups.

Our findings are consistent with observations by Halter et al. [47], where they measured the impedance of benign and cancerous prostate tissues. Gabriel et al. reported average increases in conductivity from 0.1 to 0.3 S/m over the frequency range of 1 kHz to 1 MHz for a number of different human tissue types [48].

The Bode plots along with their model fits for all 16 specimens are shown in Fig. 12 and their corresponding circuit parameters are tabulated in Table I. Our initial results indicate that increasing malignancy leads to a corresponding increase in the tissue impedance. The mean tissue resistance Rt in cancerous specimens was 7125.3 Ω compared to 434.19 Ω in benign specimens, while the mean tissue capacitance Ct was observed to be 0.016 nF in cancerous specimens compared to 1.229 nF in the benign specimens. At a frequency of 200 kHz, the mean tissue impedance increased from 236.60 Ω in benign specimens to 7052.0 Ω in cancerous specimens. Since the sample size in our present work is small (8 benign and 8 cancer specimens), we used t-test. We carried out t-test at 200 kHz for benign and cancerous tissue cores and found that the p-value was less than 0.00001667, which indicates that the difference in the impedance value is statistically significant. From Table 1, it is also evident that the fitted circuit parameters Cdl and RPBS differ in benign and cancerous breast tissue specimens. This is most likely attributed to alterations in the ionic environment in the vicinity of the breast tissue cores. The results obtained from the present work (Cdl = 0.7 nF to 10.6 nF, Ct = 0.027 nF to 2.935 nF and Rt = 86 Ω to 9734 Ω) follows similar trend as reported in literatures focusing on measuring impedance in cells [32].

Fig. 12

Plots of (a) impedance and (b) phase measurement for benign and cancerous breast tissues using microchip having IDEs with 10 µm spacing. Also overlaid are the respective fits in red (for cancerous breast tissue) and blue (for benign breast tissue).

Table 1

Fitted model parameters of the equivalent circuit for benign and cancerous breast tissue specimens shown in Fig. 12

Tissue Label	RPBS (Ω)	Cdl(nF)	Rt (Ω)	Ct (nF)	Zt (Ω) (at 200 kHz)
Benign 8023-1	222.02	1.777	314.33	0.211	313.24
8023-2	155.81	1.994	393.57	0.177	155.72
26945-3	35.349	8.281	86.564	1.480	35.272
26945-4	31.488	10.654	132.12	2.935	31.278
27122-5	29.871	8.329	284.22	1.114	264.06
27122-6	31.843	6.283	896.81	1.674	419.88
26106-7	47.029	5.288	265.48	0.450	262.54
26106	35.26	6.547	1100.5	1.796	410.86
Mean	73.583	6.144	434.19	1.229	236.60
Cancer 25313-1	851.01	1.617	5821.5	0.023	5735.9
25313-2	971.69	1.781	5710.6	0.013	5682.4
24468-3	1274.4	0.931	5706.9	0.027	5596.9
24468-4	911.39	1.359	8456.4	0.010	8400.4
24531-5	866.06	0.869	9734.5	0.011	9634.5
24531-6	1096.2	0.780	7893.5	0.011	7843.6
3343-7	1166.4	0.705	6968.9	0.018	6882
3343-8	720.7	0.933	6710.7	0.017	6641
Mean	982.231	1.121	7125.3	0.016	7052.0

Open in a separate window

Many changes occur during the onset and progression of tumor. Morphologically, cancer cells tend to be larger, with increased nuclear-to-cytoplasmic ratio and often basophilic due to heightened protein and nucleic acid synthesis. Biochemists found many disturbed signaling pathways showing either enhanced or diminished activities. Another famous and important phenomenon is that cancer cells loose contact inhibition which implies the usual anchoring and signaling between cell membranes, as well as between cells and extra-cellular matrix, have significantly altered. While it is impossible to prove any direct causation relation at the current stage, we believe that the observed impedance change reflects the overall tissue change in the cancer development.

Go to:

4. Conclusions

A microchip capable of detecting benign and cancerous breast tissue is successfully fabricated using MEMS technology. The bioimpedance measurement technique was used to analyze benign and cancerous breast tissue cores that are fixed on the microchips. We found that cancerous breast tissue specimens displayed significantly different bioimpedance characteristics compared to benign breast tissue specimens. It was also observed that by decreasing the electrode spacing, the effective electrode area is increased, thereby increasing the sensitivity of the device. This is our initial attempt to investigate the feasibility of using bioimpedance measurement to assess cancerous tissue based on tissue specimens obtained through tissue microarray technology, which extracts uniform sized tissue cylinders from paraffin-embedded tissue blocks with a hollow needle. Though significant efforts were made to ensure uniform preparation of specimens, we recognize that there may exist minor intraand inter-experimental differences in experimental conditions including tissue heterogeneity, microtome settings, environmental temperature, etc. Further works in this area with more cases, different geometries of the tissues and electrodes, and variety of disease types will allow us to examine these factors and investigate them in more detail to further understand the impedance measurement phenomenon more clearly. We also plan to integrate bioimpedance measurement with mechanical characterization for automated sampling of breast tissue core specimens and apply this technology to investigate a broader set of breast diseases or other organs, especially at different stages of cancer progression.

Go to:

Acknowledgement

This work was supported by the National Cancer Institute of the National Institutes of Health grant R01CA161375.

Go to:

Biographies

• 

Hardik J. Pandya is currently in Post-Doctoral position at the Robotics, Automation, and Medical Systems (RAMS) Laboratory in the Department of Mechanical Engineering at University of Maryland, College Park, USA. He completed his Bachelor of Science in Electronics (2002) and Master of Science in Electronics with gold medal (2004) from Sardar Patel University, Vallabh Vidhyanagar, Gujarat, India. He received his Ph.D. in Microelectronics Engineering from Instrument Design Development Centre, Indian Institute of Technology Delhi, India in 2013. He worked as Project Fellow at Department of Electronics, Sardar Patel University from June 2004 to July 2006. From August 2006 to February 2009, he served as a Lecturer in Department of Electronics, Invertis Institute of Engineering and Technology, Bareilly, U.P., India. From May 2009 to August 2012, he worked as Sr. Research Fellow and Research Associate in Center for Applied Research in Electronics, Indian Institute of Technology Delhi, India. His research interests include design and fabrication of Bio-MEMS, Flexible electronics, Bio-sensors, Microfluidic devices, synthesis of metal oxide nanostructures and their applications.

• 

Hyun Tae Kim received the Bachelor of Engineering degree in Mechanical Engineering from Korea University, Seoul, Korea, in 2006, and the Master of Science degree in Electrical Engineering from the University of Utah, Salt Lake City, USA, in 2012. He is currently working towards his Ph.D. degree in Mechanical Engineering at the University of Maryland, College Park, USA. His research interests include MEMS, robotics, automation, and medical systems.

• 

Rajarshi Roy is currently a working as a Postdoctoral Research Scholar in the Department of Mechanical Engineering at Vanderbilt University, Nashville, Tennessee. He completed his undergraduate studies from Jadavpur University, Kolkata, India in 2008 with a B. Eng degree. He received his M.S. and Ph.D. in Mechanical Engineering from the University of Maryland, College Park in 2013 and 2014 respectively. His research interests include image-guided micromanipulation, AFM based micro-scale tissue characterization and soft tissue biomechanics.

• 

Wenjin Chen received her Ph.D. degree from joint program in Molecular Biosciences at University of Medicine and Dentistry of New Jersey and Rutgers, the State University of New Jersey in 2005. She is currently working as Associate Director, Computational Imaging at Center for Biomedical Imaging & Informatics, Rutgers Cancer Institute of New Jersey, and oversees the imaging services at Histopathology and Imaging Core Facility. Her research interest focuses on utilizing image processing and computer vision methods, robotic and virtual microscopy to facilitate new technology development in cancer research.

• 

Lei Cong received the MBBS degree from China. She has both HTL and QIHC certifications from ASCP. She is the Supervisor of Histopathology and Imaging Shared Resources, and the Biospecimen Repository Services in Rutgers Cancer Institute of New Jersey, New Brunswick, New Jersey, USA. She oversees both the Histopathology and Imaging Services, and the Biospecimen Repository Services. She is responsible for quality control of tissue specimens, tissue microarrays, specimen storage and acquisition, histology and IHC, and biospecimen database managements, etc. She is engaged in many research projects involving tissue microarrays, multispectral imaging analysis, study design, clinical trials, and biomarkers.

• 

Hua Zhong studied medicine in China where he also had his initial clinical and research training. In 2001, following a 3-year fellowship at Johns Hopkins Oncology Center, he became a research faculty member at Emory University School of Medicine. Currently, he is a boardcertified surgical pathologist at Rutgers Robert Wood Johnson Medical Group, teaches medical students at Rutgers University, co-leads biorepository service and provides pathology support for research projects conducting at Rutgers Cancer Institute of New Jersey.

• 

David J. Foran earned a bachelor’s degree in Zoology and Physics from Rutgers University in 1983 and received a Ph.D. in Biomedical Engineering jointly from Robert Wood Johnson Medical School and Rutgers University in 1992. He was recruited to the Faculty and currently serves as Professor of Pathology, Laboratory Medicine & Radiology and Chief of the Division of Medical Informatics at Rutgers -Robert Wood Johnson Medical School and as Executive Director of Biomedical Informatics & Computational Imaging and CIO at Rutgers Cancer Institute of New Jersey. A major concentration for Foran’s laboratory has been the development of a family of data-mining, imaging and computational tools for characterizing a wide range of malignancies and elucidating the role that protein and molecular expression plays in disease onset and progression. This work has resulted in numerous publications, invited book chapters and several pending and issued patents. This work has received competitive extramural funding from the Whitaker Foundation, the NJ Commission on Science & Technology, the federal Defense Advanced Research Projects Agency (DARPA), the Department of Defense (DoD), the Radiological Society of North America, the National Institutes of Health (NIH) and the private sector.

• 

Jaydev P. Desai is currently a Professor in the Department of Mechanical Engineering at University of Maryland, College Park (UMCP) and the Director of the Robotics, Automation, and Medical Systems (RAMS) Laboratory. He completed his undergraduate studies from the Indian Institute of Technology, Bombay, India, in 1993 with a B.Tech degree. He received his M.A. in Mathematics in 1997, M.S. and Ph.D. in Mechanical Engineering and Applied Mechanics in 1995 and 1998 respectively, all from the University of Pennsylvania. He is a recipient of the NSF CAREER award and the Ralph R. Teetor Educational Award. He was an invited speaker at the 2011 National Academy of Sciences Distinctive Voices at The Beckman Center and was also invited to attend the National Academy of Engineering’s (NAE) 2011 U.S. Frontiers of Engineering Symposium. His research interests include image-guided surgical robotics, rehabilitation robotics, haptics, reality-based soft-tissue modeling for surgical simulation, model-based teleoperation in robot-assisted surgery, and micro-scale cell and tissue characterization. He is a member of the ASME and IEEE.

Go to:

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Ten AI Trends Revolutionising Consumer Electronics

The consumer electronics industry is constantly evolving and growing at a fast pace. It has seen a number of new products and developments in recent years, which has led to aggressive investment and innovation. Presented in this article is a roundup of the latest AI trends in the consumer electronics industry.

Sani Theo

What if your electrical home appliances become live— hear you, see you, feel your emotions and become more interactive? Some of these features have already been made available in artificial intelligence (AI)-based smart home appliances. Interactive and emotional toys, smart fridges, smart thermostats, voice assistants and others are now found in many homes.

Created by a Japanese company, mui is an interactive wood panel that serves as a smart home control hub.

There is no doubt that AI technology is already making a huge impact on our daily lives.

We have seen the capabilities and applications of AI in the manufacturing industry, in oceans and even in the air. Now, AI is gaining importance in consumer electronics as well. Traditional user interfaces in consumer devices are being replaced with abstract user-interfaces such as voice-based and gesture-based. This emerging technology has created not only many new jobs but has also changed consumer behaviour across the globe. Some experts say that instead of using cellphones, smart entertainment devices and smart home electrical appliances could be used more often with intelligent voice control in the future. While there are many ways that AI is changing our world, let us explore ten trends that are worth mentioning and that are revolutionising current consumer electronics.

1. Smart homes. The complete automated home is a new concept, but slowly the world is moving towards its true realisation. The Internet of Things (IoT) and AI are being integrated into many household products. Such AI-powered capabilities ensure convenience, comfort and affordability for users. In a smart home, intelligent devices can automatically get switched off or on, and be monitored in real time using smartphones.

2. Smart TVs. Internet-connected TVs have become a lot more smarter with AI technology. As per a recent report, the global smart TV market accounted for US$ 158.4 billion in 2018, and this number is expected to increase further in the years to come.

Next-generation intelligent smart TVs can connect to the Internet wirelessly, and access online streaming services.

AI-powered TVs can suggest movies, TV shows and Web series based on viewing habits of the customers. These have enabled browsing such social media applications as Instagram and Facebook, stream live videos and check emails, just like on smartphones. Future smart TVs will deliver even better quality pictures using less bandwidth.

3. Voice-enabled technology. We are witnessing the use of voice assistants, such as Amazon Alexa and Google Assistant, in households across the globe. These AI-powered devices also have a presence in developing countries like India. Report says that voice-activated technology is a rapidly expanding market, with over 150 million Indians likely to use voice assistants by the end of 2019.

However, a huge country like India has vast language barriers. Ninety per cent of voice assistants currently available in India only support English. Hence, Indian startups such as Liv. ai, Vokal and Reverie Language Technologies are making strides to create AI technology for consumers in regional languages. This move could enable consumers in poorer, rural communities to benefit from voice-enabled technology.

4. Instant language translation. No language barriers anymore? Well, that is true with the latest AI-powered language translation devices. Such translators are becoming popular, especially among frequent travellers and students. Although, Google Translate app can be used as a language translator, it still falls short of some languages, especially Asian ones.

There are dedicated devices that are more efficient and powerful than Google Translate. For example, Muama Enence instant translator is one such device. It enables real-time two-way communication across the globe. It can also be used to learn foreign languages. All you need to do is press the button, speak and get the voice translation in an instant.

5. AI-driven education. Robots that play games to assist human beings are being used in schools. AI technology is being used to assist children with disabilities or encourage technical creativity. It might soon be a teacher’s new best friend. As compared to humans, AI is less likely to make content-based errors. It can access large amounts of data from the network and give correct answers. It can read faces of students and respond accordingly.

6. Smart security systems. Capabilities of traditional surveillance cameras can be enhanced using AI, the IoT and machine learning. Modern smart cameras can be trained to monitor specific locations, generate alerts and take real-time actions to safeguard user’s assets. Such systems can also provide facial recognition to control access, detect motion, detect sound and connect with smart home systems. Based on specific user preferences, the cameras can send personalised alerts for specific incidents.

7. Online shopping. This is a global trend that is shaping the behaviour of consumers. AI algorithms provide product recommendations based on assessments of individual buying patterns and product preferences, making the shopping experience more relevant, enjoyable and satisfying. There is a good presence of AI technology in apparel, fashion and athleisure. These products are designed to suit consumers’ individual features and needs.

Food and beverage companies can customise their packaging using digital printing technology.

8. Evolution of consumer behaviour and adaptation. AI has embarked on its journey of revolutionising consumer electronics. Consumers are trying to embrace the opportunities brought about by AI technology. However, consumer organisations can embrace AI only if they are ready to adapt. Technology companies are racing ahead of the curve in the fast-paced AI-driven world. Digital products and services are made as inclusive and affordable as these are innovative. Regulators and developers are required to create an environment where these emerging technologies are built with consumer safety, privacy and security in mind.

9. Data protection. Governments around the world are trying to enhance data protection regulations and put the right regulatory standards in place to protect consumers against data breaches. There is a need for robust data protection laws across the world. There are many countries that have data protection laws in place. Some are leading the way by converging their domestic laws to universally recognised standards. Others have laws in place, which are not as robust as they could be but provide a good starting ground for improvements.

10. Regulators and developers. AI is growing at a fast speed while at the same time there are a number of issues to be addressed, including the danger of killer robots, digital privacy for consumers and other ethical issues. There are regulatory bodies that monitor and control various AI developments to protect consumers. However, there are conflicting opinions as to whether AI should be governed by binding law, or whether certain areas can be left to a code of ethics and self-regulation. Certainly, consumer behaviour will depend on the regulations and developers.

Conclusion

Applications of AI in consumer electronics open up many growth avenues and opportunities. The IoT, AI and machine learning are being used in many consumer electronics. Thus, devices will soon become more autonomous and get better at assisting consumers in the future. As technology progresses, consumer electronics companies are racing to provide next-generation smart products that are powered by AI technology.

Companies that can offer real values and delightful experiences to users are likely to thrive. Affordable access to digital products and services would enable consumers to tap the true potential of AI in developing countries across the globe. And we as consumers will witness the real AI explosion in consumer electronics in the country

The IoT For Environmental Monitoring: Needs And Challenges In India

This article is based on a speech given by Poonam J. Prasad, senior scientist, Analytical Instrumental Division, CSIR - National Environmental Engineering Research Institute, Nagpur, at IOTSHOW.IN 2019, held in Bengaluru. The institute’s Analytical Instrumental Division focuses on R&D on environmental sensors and the Internet of Things (IoT). Speaking on the use of the IoT for environmental monitoring, Prasad highlighted the need for IoT-based environmental sensors to conserve energy, water and other resources.

We need the Internet of Things (IoT) in environment monitoring to be able to conserve energy, water and other natural resources, which are being contaminated every second. In conventional environmental monitoring methods, samples are collected, analysed and analytical instrumentation is carried out on them.

There are two ways of doing this. One is manual, where the sample is collected and analysed in a lab. Second is instrumental, where the quantity of pollutants in the sample is analysed, on the go, automatically.

Instrumental methods have direct analytics, where readings and results are automatically received. Manual methods, on the other hand, need pretreating the sample before carrying out sedimentation, isolation and other processes on it.

When we talk about environmental monitoring using the IoT, we primarily focus on such areas as waste management, air pollution and extreme weather.

Why we need the IoT for environment monitoring

When we go deep into environment monitoring, it is a very complex system and, hence, we cannot just start using sensors for regulatory purposes. If we have data for water and air, then we can use AI and ML tools, among others. There are environmental sensors for measuring water quality, radiations and hazardous chemicals.

Similarly, in the industrial IoT (IIoT), we need methods for ensuring safety of workers, because some industries generate obnoxious gases like sulphur, methane and sulphur’s compounds, which are bad for human health. By getting data out of sensors, we can maintain a good safety record. Places that are inaccessible can also effectively utilise sensors.

Since 2012, research is happening all over the world on environmental sensors. Some reviews have already been done. People have done outdoor air-quality monitoring using a ZigBee-based wireless sensor network. However, indoor environment is generally more polluted than outdoor environment and, hence, the system developed for the outdoor environment may not suit indoor environment.

Researchers have developed an air-quality system that records particulate matter (PM). The system categorises PM into PM 10, PM 2.5 and PM 1. Once PM goes into the lungs, it leads to health issues. Standards are being implemented for PM 1, but more precisely PM 10 and PM 2.5 are being monitored.

Mobile sensing systems have been developed and proposed for recording PM 2.5 in cities. Some research papers have described low-cost, portable monitoring systems, which monitor multiple parameters such as humidity, PM 2.5, volatile organic compounds (VOCs), CO2, CO illuminance and sound levels.

Sensors are divided into two categories: electrochemical-based and metal-oxide-based. Companies use these sensors based on their requirements. Both types of sensors have advantages and disadvantages. But research is being done mostly on metal-oxide sensors to get more sound results for environmental monitoring.

Likewise, a micro sensor-based air quality monitoring system has been developed for real-time monitoring of airborne, fine particulates. It has already been tested.

Top sensors used in the environment

As the environment is heterogeneous, the system needs to be utilised well, because we cannot develop one protocol-based system and expect it to work in all situations. Therefore we need a multi-protocol system. Also, it is important to understand the interference of pollutants, because pollutants such as ozone, NO 2 or NOx particles have interference capability. Therefore the science behind this interference, how data is coming and what could be the reason for any deviation in data must be studied and understood. Only then can a sensor be well-characterised and developed.

Top sensors used in the environment are:

• Temperature sensors

• Proximity sensors

• Water quality sensors, which measure pH, BOD, COD and other microbial contaminants; these also measures ion parameters like arsenic, iron or other compounds

• Gas sensors, which detect air quality conditions

• Smoke sensors, which are required for industrial environmental conditions or smoke-prone places

US Environmental Protection Agency (USEPA) has evaluated sensors using conventional methods so that these can be utilised for research purposes and IoT applications. Alphasense OPC N2 sensor is for PM 10 and PM 2.5 monitoring. This was tested through GRIMM, which is a certified handheld monitor. So far, these sensors have not been internationally certified. Essentially, these are not USEPA-certified sensors, but are USEPA-evaluated sensors. This is because the technology is new, and it keeps on evolving. Every six months there is a new version of these sensors.

AQMesh, CairClip and CitySense are gas phase sensors. These are being evaluated by USEPA, and are internationally-funded projects. The systems are being tested against standard instrumentation techniques.

A typical regulatory monitor is quite expensive, and is based on analytical methods (not sensor). It is highly-reliable, but stationary. Moreover, trained staff is required to operate it. One of its advantages is that it can operate for more than ten years. But it needs to be calibrated quarterly.

On the other hand, a typical low-cost monitor does not require too much training, but then it has a limited lifetime.

Challenges in deploying IoT-based sensors

The current technology is expensive, provides only a snapshot of data, requires expertise to use and takes time in lab analysis.

National Environmental Engineering Research Institute is developing a new technology that includes the IoT, and will be low-cost, easy to use and provide continuous data. However, such technology needs to have a QA/QC approval, and there is no common agency for approving these techniques.

Major research findings for sensors or systems have been in microprocessors. The system being developed at National Environmental Engineering Research Institute (NEERI) includes a wide variety of low-cost components (varying from US$ 100 to US$ 300).

Also, if you are not a good integrator, you cannot integrate these components well. You also cannot use multiple sensors in a single board. However, if a balance can be maintained between power, cost and latency, the system can be used in the real environment.

Sensor characteristics include stability, detection limit, repeatability and reproducibility, and cost, while user requirements include measurement duration, data quality and budget.

A sensor’s lifetime is only two or three years. Sensitivity, stability and longevity of the sensor need to be improved for its operation.

The Indian government has defined air quality index (AQI). It uses one number, one colour, one description to judge air quality. From this, it can be known that PM 10 is the highest polluting among all pollutants.

Council of Scientific & Industrial Research (CSIR) has conducted its own case study by installing ten IoTbased sensors in Delhi. The findings say that low-cost sensors dominate the market and a few sensing elements exist. More research needs to be done on sensing elements. PM sensors are widely available as compared to gas phase sensors.

Two major challenges for sensor application are:

• Sensor performance values vary widely.

• Basic testing by manufacturers is lagging.

Since 2012, there has been a huge cost reduction. Reliability is there but more is required before getting to reality. Also, the cost involved in installation, maintenance and data analysis needs to be reduced. Going forward, we are looking towards seamless implementation, data quality and reliability.