- Session 1
- Session 2
- Session 3
- Session 4
- Session 5
- Session 6
"Architectures, low-power radio and signal processing in a converged sport & fitness environment"
Niclas Granqvist, Technology R&D, Polar Electro Europe Bv
With Internet in cell phones and the soon to be published Bluetooth Low Energy specification the personal area network (PAN) will converge with the mobile phone. Heart rate monitoring in sports and fitness is studied as a use case throughout. We give an introduction of all the major components of the architecture, in particular IEEE-11073 and the Sensor Profile as ways of achieving interoperability. In the end we look at the heart rate detection problem and give an overview of the signal processing challenges.
“Flexible electrode devices for stimulation and recording”
John Alderman, Tyndall National Institute
This talk will outline of some of the electrodes that have been fabricated at the Tyndall for retina stimulation and peripheral nerve recording and stimulation. These are based on a ultra thin polyimide technology with noble metal electrodes and a novel delamination technique to release the devices. The present project to add local amplification / noise reduction for cuff electrodes for peripheral nerve recording will also be discussed.
"Power reduction using truncated multipliers"
Manuel de la Guia, University of Limerick
Digital signal processing applications need complex arithmetic functions for implementing filters and real time statistical operations. Power consumption in these systems directly relies on the multipliers precision. On the other hand, the resolution of the signal processing block also relies on the multiplier precision, where lowering the bitwidth will result in less accuracy. This dilemma becomes specially problematic in portable devices where the optimum equilibrium needs to be found in order to get acceptable results with the longest possible battery life. A configurable approach where the multiplier bitwidth can be modified at runtime will be detailed. It allows the accuracy of the output to be dynamically reduced or extended, in order to get power-related reduction, as the dynamic power consumption can be optimized to the application needs.
"Wireless Communication with a deeply implanted blood pressure monitor"
Olive Murphy, Institute of Biomedical Engineering, Imperial College London
Surface acoustic wave devices have become ubiquitous in mobile communications due to their size and high stability. The inherent properties of these piezoelectric devices have been further developed for use as wireless tyre pressure monitors found in high-end cars and large trucks. More recently the same electromechanical characteristics have been tuned for use as implantable blood pressure monitors to help provide clinicians with real-time, accurate data on a 24/7 basis. The transition from tyre pressure monitor to deeply implanted blood pressure monitor is non-trivial for many reasons. One of the biggest obstacles surrounds the wireless communication with such a deeply implanted sensor. The important topics of frequency selection versus power budget and aerial design are presented. The use of bio-phantoms for in-vitro testing along with in-vivo testing in pig hearts will also be presented along with a general overview of the sensor and signal processing techniques.
Go to the top
"Low power and low complexity fixed and flexible DSP"
Oscar Gustafsson, Linköping University
For DSP functions having fixed functionality it is possible to simplify the corresponding implementation significantly to reduce the power consumption and computational complexity. In this talk we will present different ways to obtain this. Primarily, we will discuss how to replace costly general multipliers with simpler networks consisting of adders, subtracters, and shifts and show how these methods can be generalised to multiple-input non-recursive time-invariant linear transfer functions. Furthermore, we will present algorithms with a certain degree of flexibility that are composed of both fixed function blocks and simple reconfiguration blocks to reduce the total implementation cost.
"DSP and ultra wide band radar for early breast cancer detection"
Martin O'Halloran, National University of Ireland, Galway
Ultra Wide Band (UWB) radar, combined with DSP-based beam-forming and image processing, has been proposed as an alternative imaging modality to X-ray mammography for the early detection of breast cancer. Its attractions include low power operation, the use of non-ionising radiation, and the potential for good tumour localisation. This talk will cover the issues associated with the use of UWB for this application, will discuss current research and potential performance, and will highlight current challenges and future work.
"Software implementation of concatenated Huffman and Reed-Solomon encoder/decode for Medical WSN application"
Richard McSweeney, University College Cork
This talk is about the system level power reduction of Wireless Sensor Nodes in medical applications by the use of source and channel coding. The reliability and energy savings of the serially concatenated Huffman and Reed-Solomon(28,24) codes is presented, and the scheme with regards to medical applications in patient monitoring is discussed. The system was implemented on the Tyndall 25mm motes (Atmel mega128 processor and Nordic nrf2401 transceiver) that operate on the 2.4 GHz ISM band. Sample EEG data was fed into the system, and encoding decoding procedure was performed. The results show that it is possible to have BER and compression gains in the system, but also show that computational time of purely software oriented implementations are not optimal.
Go to the top
"CMOS-based microsystems for biomedical applications"
Diego Barrettino, University College Cork
Research activity on microsystems is currently directed towards single-chip solutions with micromachined devices which are becoming more and more complex, and system specifications which are more demanding. At the same time, the price of the microsystems and the time to market must be reduced. Even though the current improvements in micromachined sensors and actuators performance are amazing, they do not necessarily meet the demanding microsystem specifications for biomedical applications. Sophisticated readout and control electronics have to be designed and monolithically implemented with the micromachined devices to respond to such challenges. The aforementioned aspects will be illustrated in this lecture by presenting some examples of microsystems comprising micro-electro-mechanical devices (membranes or cantilevers) and associated readout and control circuitry.
"Computer aided diagnosis in thoracic CT"
Keelin Murphy, University Medical Centre, Utrecht
In recent years the amount of medical image data requiring expert analysis has increased significantly. Diagnostic scans including 3D and 4D datasets are becoming ever more prevalent and image quality and resolution are constantly improving. The analysis of all this data is a time-consuming and error-prone task for human experts. This talk will focus on the development of automatic image processing algorithms to assist radiologists in the interpretation of chest CT scans.
"Comprehensive sampling for biomedical applications an introduction"
Brian McGinley, National University of Ireland, Galway
A central objective of EEDSP is the efficient representation of biomedical signals. The traditional approach to acquiring and representing a signal efficiently is through Nyquist sampling followed by either direct or transform-based compression. This methodology is often wasteful as much of the data acquired via sampling is discarded at the compression stage. Compressive Sampling is an emerging field in DSP where such wasteful sampling is eliminated by simultaneously acquiring and expressing the signal in a sparse basis. This talk provides an introduction to compressive sampling and relates the concepts involved to traditional compression methods.
Go to the top
"Ambulatory monitoring in exercise and sport"
Alan Donnely, University of Limerick
Recording Physical Activity (PA) levels is becoming increasingly important in public health and in exercise science. Physical Activity is a major contributor to energy expenditure, and lack of PA is a key factor in the increasing obesity levels seen in most developed nations. To obtain a reasonable estimate of PA in any individual, a sample period of 5-7 days or longer is necessary, and any device used needs to be both comfortable to wear and capable of recording detailed information over this period. Existing commercially available devices use recording accelerometry, ECG, heat flow or a combination of these factors to estimate PA and energy expenditure. Factors that influence accuracy and functionality of recording devices will be discussed.EMG is widely used in strength training research and in rehabilitation to provide feedback on muscle activation. However, EMG MPF and MMG are less commonly employed. EMG MPF in particular has the potential to provide good information on muscle training adaptations and on muscle fatigue. Most muscle function studies to date have been lab based, and there is little information on MPF or MMG changes during training and competition. The potential of these measurements in sport will be discussed.
"A wireless Communication unit for biomedical implant applications"
Colm Mc Caffrey, Tyndall National Institute
One of the most fast-growing applications of electronic microsystems has been seen in the biomedical area, where these systems incorporate different sensors and actuators, which together have the potential to provide early–disease detection, monitoring and treatment. The essential component of such a system, in particular for implanted applications, is a wireless communication unit that delivers data from the patient to the clinician, and/or feedback from the clinician to the device which, if needed, can activate actuators to perform certain tasks in-vivo. The presented study describes results of the development of a low power highly miniaturised wireless transmission system that is applicable for implantable operation. The system operates in the 433.92 MHz and uses a stripline loop antenna with a mean diameter of 10mm and a track dimension of 2mm width by 35um thickness. Stripline loop antenna provides great advantages in terms of miniaturisation; however there is a great cost in terms of antenna impedance sensitivity. This makes the antenna difficult to match, with implications for power delivery efficiency and effectiveness of the communication system. The study presents the theoretical approach to the matching of the stripline loop antenna. The effect of component value errors was explored, as was the effect of changing impedance due to the dielectric in the near field. A simple solution to the impedance sensitivity problem was suggested with a shunt resistor used across the antenna to degrade the quality factor. The results of the experimental investigation of the antenna compared with theoretical results are presented.
"Artefact detection in EEG"
Simon O'Regan, University College Cork
One of the main obstacles in creating effective artefact detection algorithms is the lack of access to an extensive, annotated artefact database. This talk describes the collection of over 300 minutes of EEG artefacts from 20 participants. The talk will then briefly describe some current work on SVM-based artefact detection classifiers.
Go to the top
"Low power processor platform for biomedical signal processing"
Kunjan Patel, University College Dublin
There are various biosignals used by clinicians for diagnosis of various diseases. These kinds of signals are generally one dimensional and hence their analysis is different from the image analysis in biomedical signal processing. Although some of the signals can be sampled and analyzed at comparatively low frequency rate, usually multi-channel biosignals are used for the diagnosis purpose. To make the biomedical devices portable, they are made battery operated and expected to last long. Therefore amount of signal processing should be increased in the system to save power. Because of advances in sensor technology highspeed sensors are now available and hence to match the speed of these sensors low power multi-channel ASICs have been developed which can collect the data at very high speed. So, an architecture which can process this data in real time is required. We propose an approach to process biosignals efficiently in terms of speed and power. The approach is based on Single Instruction Multiple Data (SIMD) approach and enhanced for low power signal processing. The architecture was derived and tested for some basic biomedical using Java Instruction Set Simulator (JISS).
"Energy efficient protocol for reliable data transfer in wireless body area networks"
Stevan Marinkovic, University College Cork
Energy-efficient MAC protocol suitable for communication in a Wireless Body Area Network for remotely monitoring of physiological signals such as EEG and ECG will be presented. The protocol takes advantage of the relatively fixed nature of the Body Area Networks to implement the effective TDMA strategy with very little amount of overhead and almost no idle listening. The main goal is to develop energy efficient and reliable communication protocol to support streaming of large amount of data. TDMA synchronization problems will be discussed and solutions will be presented. Equations for duty cycle calculation are also derived for power consumption and battery life predictions. The power consumption model is validated through measurements. Results show that protocol is energy efficient for streaming as well as for short bursts of data, thus can be used for different types of physiological signals with different sample rates.
"The Tyndall health monitoring unit"
Brian McCarthy, Tyndall National Institute
The aim of this work is to develop both the hardware and software systems for a Health Monitoring system that will wirelessly monitor the required parameters which will enable health care practitioners to observe the activities and physiological response to patient activity. The target market for this system is for use in Ambient Assisted Living for elderly clients who require the system to be wearable (flexible), unobtrusive (highly miniaturised), have long life (reliable and low power consumption). It is also important that the system be implemented in a wireless fashion so it can provide the utmost comfort for the client and assure them and their families of their safety. The Health Monitoring Unit to be presented comprises an Electrocardiograph for monitoring Heart Rate, a Pulse Oximeter to monitor the oxygenation of haemoglobin and an accelerometer for eliminating motion artefacts which are prominent in both the ECG and Pulse Oximeter. These are especially important when these functions are implemented as part of a wearable system. The accelerometer can also be used to monitor falls and to determine the general orientation of the patient. A temperature sensor has also been included to reduce artefacts that occur due to changes in temperature. All of this data is transmitted to a remote base station through Zigbee communications for further data processing. Basic processing is available onboard 25mm mote at the base station and if need be, an FPGA layer can be added to the receiver to implement high end signal processing. It is also possible for the base station to actuate an emergency response when appropriate. This Health Monitoring device has been developed on Tyndall’s 25mm prototyping mote1. However, the final application of this unit will be implemented on either a 10mm platform or else on flexible PCB, both of which can be used as part of a wearable system.
"Classifiers and controllers implementation using an evolved hardware spiking neural network"
Seamus Cawley, National University of Ireland, Galway
This presentation describes a reconfigurable Network on Chip (NoC)-based Spiking Neural Network (SNN) architecture and FPGA implementation. A Genetic Algorithm-based evolution framework evolves SNN-based solutions to controller and classifier problems. Application of the SNN approach to biomedical-related problems is proposed.
Go to the top
"Sensing technology for gerontology tele-health: perspectives"
Gearóid Ó Laighin, National University of Ireland, Galway
The dramatic growth in the proportion of older people in communities worldwide is well established. This growth will require increase use of tele-health to manage the health status of older people. This talk will reflect on experience in developing sensing devices for older people for use in a home setting, with particular focus on mobility monitoring and fall detection in the elderly.
“Evaluation of the use of mechanomyography (MMG) with accelerometers for detecting muscle fatigue”
Anthony O’Brien, University of Limerick
Surface Electromyography (sEMG) is a non-invasive method employed for detecting muscle activity by measuring electrical activity. Mechanomyography (MMG) offers an alternative approach, whereby mechanical attributes can be measured. This talk will present the details of a recent study into using MMG for the detection of fatigue of the biceps brachii. The experimental setup will be described and details of the resulting measurements presented. Existing methods and algorithms for analysing the measurements will be discussed, as well as the results of the study.
“Seizure detection in the newborn”
Nathan Stevenson, University College Cork
The presences of seizures in the newborn are a clear sign of central nervous system dysfunction. The diagnosis or detection of the presence of seizures is, therefore, an important clinical problem. Currently, the most effective tool for determining seizures is the EEG as clinical manifestations of seizure are typically suppressed in the newborn. The interpretation of the EEG is, however, difficult, and can only be performed by highly trained personnel who are not always accessible to the clinician. This has provided an opportunity for automated methods to be applied to EEG analysis. This presentation begins by detailing the signal characteristics of the seizure and nonseizure waveforms seen in newborn EEG, outlines several methods that have been proposed for detecting seizure (including the latest algorithm developed here at the University College Cork) and discusses the ramifications of applying automated decision making to a clinical setting.