Please view our lab openings for up-to-date information on research projects


(Credit for the past work highlighted here goes to our entire team of current and past graduate students and postdoctoral fellows!)


Research goal

The main goal of our research program is to help maintain and improve health by providing medical doctors, biomedical researchers and/or patients with novel microelectronic technologies for miniature medical devices that directly or indirectly interface with the human body in order to monitor its function and, in some cases, influence it. We also target scientific, industrial, consumer and environmental sensory electronics applications.




Modern healthcare practices suggest that patient-interfacing medical devices of the future are to be potent, ubiquitous, and inexpensive. Our research investigates such medical devices. These devices address specific unmet healthcare needs, particularly those in medical monitoring, diagnostics and therapy in clinics, biomedical research labs and at home. Of our immediate interest are applications in neuroscience and molecular biology. We target disorders and diseases with limited conventional treatment options or with costly diagnostics options. Specific medical applications include electronic therapy for intractable epilepsy [J21, C51, C56, J13] and electronic screening for early detection of certain types of cancer [C45, C50, J16, J27, J28, J11, J15, J24].




Interfacing with the human body for the purpose of maintaining or improving health requires a variety of sensory functionalities. These can be as simple as monitoring key vital signs, or as complex as monitoring electrochemical activity of the brain or examining biochemical content of bodily fluids. In our research we target applications where novel implantable, wearable or disposable biomedical devices with complex sensory functions are uniquely enabled by low-cost integrated circuit (IC) technologies such as CMOS.


Specifically, we focus on the design of integrated circuits, VLSI architectures and signal-processing algorithms that comprise the core of a sensory medical device.  Such sensory devices not only acquire raw sensory data, but also perform local sensory signal processing (such as feature extraction and machine learning data classification algorithms), and provide feedback information or, in some cases, feedback action as shown in Figure 1. One successful example of such a system is a single-chip brain implant for treatment of drug-resistant epilepsy we developed, that accurately anticipates seizures and automatically triggers neuro-stimulation to effectively abort them [J21, J29, J30, J31, C51, C56].


Screen Shot 2017-02-04 at 8.44.22 PM              Screen Shot 2017-02-05 at 10.26.26 AM      


Figure 1. Functional block diagram of a biomedical sensory microsystem (left) and an illustration of its typical form factor (right).





1. Sensor/Actuator Fabrication


From the sensor/actuator fabrication and microsystem integration perspective, for potency, ubiquity and low cost it is often advantageous to utilize sensory properties of either the integrated circuits themselves or of additional small sensors with a similar form factor. This approach avoids bulky externally connected sensors or associated packaging costs. In our previous work, we have demonstrated suitability of silicon integrated circuits (ICs) to be further (post-CMOS) integrated with various arrays of on-die and off-die sensors for implantable, wearable and disposable microsystems implementations.


1.1 In electrophysiological sensing/actuation applications, we have developed various such prototypes including: arrays of gold micro-needles to monitor spatial maps of electrical neural activity in the brain for in vitro epileptic seizure propagation studies (Fig. 2a), active neural probes comprised of 2D arrays of platinum micro-needles directly co-integrated with neural amplifiers on an integrated circuit (Fig. 2b), and 3D arrays of both rigid and flexible microelectrodes for in vivo implantation (Fig. 2c).



 Screen Shot 2015-12-11 at 12.21.47 PM.png Screen Shot 2015-12-11 at 12.21.47 PM.png    



Screen Shot 2017-02-02 at 3.04.24 PM     Screen Shot 2017-02-02 at 3.02.45 PM   





Figure 2. Examples of sensor fabrication and microsystem integration techniques for electrophysiological sensing/actuation applications: (a) micro-needle arrays for in vitro use (with Prof. Peter Carlen) [J7], (b) active neural probe for acute in vivo use (with Prof. Peter Carlen) [C27, J9, J28], and (c) in vivo chronically implantable microelectrode arrays (with Prof. Raafat Mansour and Dr. Salam Gabran) [J22, J25, J27].



1.2 In electrochemical sensing/actuation applications, the sensors prototyped by us range from arrays of flat gold microelectrodes accessed by on-chip microfluidic structures (Fig. 3a), to nanostructured gold electrodes with affinity-based chemical functionalization (Fig. 3b), to post-implantation fouling-resistant on-gold chemical coatings (Fig. 3c). Applications include molecular diagnostics such as measuring pathogen DNA concentration and cancer screening (Fig. 3b) and measurement of concentration of neurochemicals for brain neurochemistry studies and diagnostics (Fig. 3c).




 Screen Shot 2015-12-11 at 12.21.47 PM.png  Screen Shot 2015-12-11 at 12.21.47 PM.png  isscc-fig7  isscc-fig7






 Screen Shot 2017-02-02 at 3.53.58 PM  Screen Shot 2017-02-02 at 3.58.01 PM



Figure 3. Examples of sensor fabrication for electrochemical sensing/actuation applications: (a) flat gold microelectrode arrays for cell culture monitoring in on-chip microfluidic structures (with Prof. Guenther) [J17], (b) nanostructured gold electrodes with electrostatically controlled texture, functionalized to sense target DNA concentration (with Profs. Ted Sargent and Shana Kelley) [C50, J26], and (c) on-gold chemical sensor that monitors the flow of ions in neuronal cells membranes but resist post-implantation bio-fouling (with Profs. Peter Carlen and Michael Thompson) [J35].



1.3 In electro-optical sensing/actuation applications, we have prototyped various CMOS imager sensors for both conventional non-contact and emerging contact imaging applications. These include photodetector arrays for electro-chemi-luminescence excitation and sensing (Fig. 4a) with fluidic samples delivered through microfluidic structures (Fig. 4b) as well as single-color (Fig. 4c) and multi-color (Fig. 4d) fluorescence contact imagers. Applications range from pesticide detection (Figs. 4a and 4b) to optical imaging of various micro-scale biological objects such as fluorescently labeled DNA microarrays (Figs. 4c and 4d).










Screen Shot 2017-02-02 at 4.22.45 PMScreen Shot 2017-02-02 at 4.28.27 PM Screen Shot 2017-02-02 at 4.26.19 PM



Figure 4. Examples of sensor fabrication and microsystem integration techniques for electro-optical sensing/actuation applications: (a) interleaved photodetector-electrode arrays for electro-chemi-luminescence (ECL) sensing and excitation [J15] with (b) fluidic samples delivered through microfluidic structures to the image sensor surface (with Prof. Axel Guenther) [J15]; (c) single-color (with Prof. Glenn Gulak) [J11] and (d) multi-color (with Prof. Ulli Krull) [J18, J23, J24] fluorescence contact imagers.




2. Front end: Sensory/Actuating Circuits


From the front-end circuits perspective, the key challenges are low signal-to-noise ratio, large sensory signal offset and drift, high interference levels, intrinsic electronic noise, time-varying signal source properties, various artifacts and numerous other sensory interface-related issues. In our previous work, we have addressed these issues individually by various integrated circuit design solutions (e.g., novel analog signal processing circuits) [J6, J10, J16, J19, J20, J33]. Our latest system-level projects using such circuits include electro-physiological sensors for brain activity monitoring and modulation (Fig. 5a), electro-chemical sensors for in vivo neurochemistry monitoring and in situ molecular diagnostics (Fig. 5b), as well as optical sensors for molecular detection, cellular imaging and computational photography (Fig. 5c).













Figure 5. Examples of our sensory front-end integrated circuits designs for (a) electro-physiological [J7, C27, J9, J12, J20, J28], (b) electro-chemical [J5, J17, C50, J26] and (c) opto-electronic sensory applications [J6, J10, J15, J18, J23, J24].



3. Back End: Computational Circuits


From the back-end circuits perspective, the key challenges are the ever-growing requirements for intelligent ways to process large amounts of sensory signal information (or big data), higher sensory data processing throughput, and higher integration density with a limited power budget. The power budget is often constrained by heat dissipation (such as that into the tissue surrounding an electronic implant). In our previous work, we have developed a number of circuit design techniques that break the conflicting throughput-area-power trade-offs. These include energy-efficient information-to-digital converter architectures that perform computationally-expensive feature extraction and data classification [J6, J10, J16], such as in the application of triggering therapeutic closed-loop neurostimulation (Fig. 6a), and energy-efficient mixed-signal VLSI architectures for accelerating machine learning in sensory data classification applications (Fig. 6b). Within this thrust, we are currently pursuing research on energy-efficient digital and mixed-signal VLSI accelerators for high-performance machine learning and artificial intelligence applications and computing architectures co-integrated with sensory arrays.










Figure 6. Examples of our computational back-end integrated circuits: (a) mixed-signal multi-core DSPs within closed-loop responsive neurostimulators for treating neurological disorders [J20, J21, J30, C67], and (b) implantable mixed-signal machine learning accelerators (including the world’s first support vector machine in silicon, with Prof. Gert Cauwenberghs) [J1, J2, J8, J13, J14].



4. Wireless Communication and Powering Circuits


Sensory microsystems often require wire-free and battery-free operation under strict constraints of low form factor, high data rate, high energy efficiency and low specific absorption rate. Our solutions to these challenges include low-power custom FSK and UWB radio-frequency transceivers (Fig. 7a), as well as wireless energy transfer circuits for neural recording and neurostimulation with off-chip (Fig 7b) and on-chip (Fig. 7c) RFID-type inductive power/data receiver coils [C55].




Figure 7. Examples of our RF transceivers and inductive energy transfer microsystems prototypes: (left) custom radio-frequency transceivers [J20, J21, C50, J26, C60], (middle) inductively powered wireless neural recording and neurostimulation microsystem [C55, J32, C65, J38], and (right) inductive power receiving circuits with an on-chip coil.


We always look forward to having talented graduate students from the University of Toronto and around the world join our team!