My research revolves around neural interfacing and its incredible potential for delivering life changing benefit to millions of people worldwide. In my work I have developed: a wireless implantable peripheral neural stimulation and recording system; a miniature FPGA headstage capable of neural recording, real-time spike sorting and long duration data logging; a new method of designing electrical neural stimulation waveforms; and am currently working on a miniature device capable of wirelessly monitoring all the body's vital signs (pulse, ECG, temperature, breathing, blood oxygen as well as EEG) that can be comfortably worn in your ear.
Zero Mean Waveforms
Neural stimulation has for the last 5 decades typically used a biphasic current waveform that is charge balanced - pushing a packet of charge in one direction through electrodes & tissue before reversing the direction of flow and pushing that same packet of charge back. This is often claimed to leave the electrodes electrically neutral with the tissue and minimise damaging faradaic reactions at the electrode-tissue boundary, unfortunately neither of these claims is accurate and this has been known for years.
Despite this awareness, significant effort continues to be expended on enhancing the biphasic charge balance performance of neural stimulators. My research has led me to believe that a better approach would be to use a 3 (or more) phase waveform that gives zero average voltage across the electrodes - a Zero Mean Waveform. It should be noted that this technique is not limited to neural stimulation but has applicability anywhere where alternating current waveforms are used.
The Senseback project is aimed at developing technology to provide artificial sensation for prosthetic limb users. My role was the development of the electronics - creating a peripheral neural stimulation and recording system.
The core of the system is a custom designed bidirectional neural interface chip (ASIC) with 32 channels of high voltage stimulation and recording. The chip provides a variety of neural recording gain settings - enabling it to record not just neural signals but also muscle biosignals (EMG) - and is also capable of spike detection and windowing to reduce neural recording data rates by ~2 orders of magnitude.
The chip has been fitted to a single implantable and flexible PCB that has 2 ends - a wireless link end that houses a bidirectional Bluetooth Low Energy data link & a wireless power link to an external backpack, and a neural interface end which houses the ASIC and support circuitry. This PCB is then coated with USP VI medical grade silicone to protect the electronics when implanted.
The first generation of wearable devices mostly consisted of watches fitted with accelerometers and a PhotoPlethysmoGram (PPG) sensor. However, a loosely fitting watch is a challenging setup for biosignal/body motion monitoring - it doesn't provide the stable coupling a PPG sensor ideally needs and the motion of the forearm in a wide range of activities as well as its highly variable position and orientation relative to your core organs is challenging for the accelerometer and other biosignal measurements.
The Signal Processing Group at Imperial College have for the best part of a decade been working on developing technology and algorithms that I believe will form the core of the next generation of wearable devices. These miniature ear worn devices will record all your body's vital signs (ECG, respiration rate, pulse, blood oxygen and temperature) as well as brain wave patterns (EEG). These devices will have incredible potential for health/wellbeing monitoring or contextual cueing and could offer a step change in key areas of medical diagnostic capability.