CMOS multimodal sensor based lab-on-a-chip system for personalized bio-imaging diagnosis

• Main Author: Huang, Xiwei
• Other Authors: Yu Hao
• Format: Theses and Dissertations
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/62261
• Institution: Nanyang Technological University

The world population is rapidly ageing with proportion of people aged 60-year-old and over, growing faster than any other age groups. Considering the current aging society, there is an emerging need to develop the future diagnosis with portable biomedical devices by bio-instrument miniaturization. The recent development of lab-on-a-chip (LOC) technology has provided a promising integration platform of microfluidic channels, microelectromechanical systems (MEMS), and sensors, which allow non-invasive and near-field sensing functions. The standard complimentary metal-oxide semiconductor (CMOS) process allows a low-cost system-on-chip solution to integrate sensors from multiple domains, which has raised many new design challenges. In this thesis, we have particularly studied the CMOS multimodal sensor for LOC integrated bio-imaging diagnosis system, including: 1) CMOS (capacitive-micromachined-ultrasonic-transducer) CMUT sensor for non-invasive ultrasound imaging towards the glaucoma diagnosis; 2) CMOS (ion-sensitive-field-effect-transistor) ISFET sensor for ion imaging towards the DNA sequencing application; and 3) CMOS optical sensor for microfluidic contact imaging towards the cell detection, recognition, and counting application. We will illustrate the need and application of the three corresponding bio-imaging diagnosis methods as well as design problems addressed when being miniaturized, which can be summarized as follows. Firstly, we illustrate one device-level design work using the example of ultrasound imaging. A two-channel analog front-end (AFE) IC for interfacing multi-channel high frequency CMUT array is developed with three-dimensional high resolution imaging capability for glaucoma diagnosis. The main challenge is the process integration between MEMS array and CMOS readout circuit, where flip-chip bonding is deployed. With the use of 30V high-voltage 0.18μm Bipolar-CMOS-DMOS (BCD) technology, the proposed AFE IC cell is designed to consist of two high-voltage (HV) pulsers in the transmit path, and a shared single low-noise pre-amplifier in the receiver path for area reduction. The electrical functionality of the proposed AFE IC is characterized in which the HV pulser generates a delay of 16.2ns between the 33ns input trigger pulse and the HV output pulse while driving the load capacitance of 43pF from 0 to 30V. And the low-noise preamplifier achieves over 60dBΩ transimpedance gain with 27.5pA/sqrt(Hz) input refereed noise current at 35MHz. A successful pulse-echo acoustic testing is also demonstrated with the developed AFE IC that integrates the CMUT sample in an oil-immersed environment. Secondly, we discuss one circuit-level design work using the example of ion imaging. A 64×64 1200fps dual-mode CMOS ion-image sensor is demonstrated with suppressed fixed-pattern-noise (FPN) for accurate high-throughput DNA sequencing. The main challenge of the traditional ISFET-based ion imaging is lack of faulty pH value detection. In this work, we show the solution by pruning sensed data with reference from multi-domain. A dual-mode ISFET sensor is developed, including pH sensing from chemical domain as well as image sensing from optical domain. An ISFET with standard 4T-CMOS image sensor (CIS) pixel structure is proposed and fabricated in standard 0.18μm 1P6M CIS process. After addressing physical locations of DNA slices determined by the optical contact imaging, local pH value of one DNA slice can be mapped to its physical address with the accurate correlation, which can significantly improve the DNA sequencing accuracy. Moreover, pixel-to-pixel ISFET threshold voltage mismatch or FPN is reduced by a correlated double sampling (CDS) readout circuit structure that supports both image and pH modes for large-array and high-throughput application. Measurement results show a sensitivity of 103.8mV/pH and FPN reduction from 4% to 0.3% with a readout speed of 1200fps. Lastly, we present one system-level design work using the example of microfluidic contact imaging. A microfluidic contact imaging system has been developed with poly-dimethylsiloxane (PDMS) microfluidic channel integrated on top of CMOS image sensor for flowing cell detection, recognition and counting. The main challenge of such a lensless microfluidic system is how to improve spatial resolution because of no optical lens. To resolve the raw spatial resolution limitation from pixel size, an extreme-learning-machine based single-frame super-resolution processing (ELM-SR) is proposed that can recover high-frequency loss in detected cell contacting images such that flowing cells can be still distinguished for counting. The prototyped lensless microfluidic system obtains less than 8% counting error for absolute number of microbeads; and 0.10 coefficient of variation for cell-ratio measurement of mixed RBC and HepG2 cells in solution. In this thesis, we have shown a thorough study to explore multimodal CMOS sensors in LOC towards the portable personalized bio-imaging diagnosis system, which could pave the way towards a variety of personalized diagnosis applications such as: 1) CMOS ultrasound sensor for non-invasive human body scanning; 2) CMOS dual-mode ion sensor for portable DNA sequencing; and 3) CMOS contact imaging sensor for point-of-care blood cell tests. Note that the primary novelty of this thesis is the design of CMOS ISFET ion-image sensor (published in IEEE Symposium on VLSI Circuits 2014). As a conclusion, the CMOS multimodal sensor based LOC system has been shown with the great potential to provide the future personalized e-healthcare solution for the coming aging society.