Embedded systems in medical devices

1
Embedded systems in medical devices

• Tomasz J.Petelenz, PhD
• Sarcos Research Corporation, Salt Lake City, UT
• S.C.Jacobsen, PHD
• Sarcos Research Corporation, Salt Lake City, UT

2
BackgroundSarcos Products Projects

• Medicine
• Artificial Arms
• Micro Catheters LIS
• Programmable Drug Pumps
• Medical Monitoring
• Diabetic Catheters
• Artificial Kidneys
• Home IV Systems

• Micro Systems
• Sensors
• Strain (1,2,6 axes)
• Pressure Flow
• Vibration, Sound Accel.
• Multi Axis Fluid Shear (skin friction)
• Rotary Encoder
• Actuators
• Servo Valves - Micro Motors

• Entertainment
• Jurassic Park Robots - large
• Disney Robots
• Ballys and Primadonna Robots
• Ford Tele Entertainer
• Carnegie Science Museum
• Buffalo Bills - Zinger
• Bellagios Robot Fountains

• Robotics Tele-robotics
• Utah/MIT Dexterous Hand Master
• Dexterous Arm I GRLA Master
• NASA Space Suit Tester
• Sensor Suit - Motion Capture
• VR Mobility Ports - IPORT
• Wireless Communication
• Other Human-Interactive Systems

3
Embedded processors in medical devices

• Already ubiquitous growing at the rate exceeding
  that of desktop computing
• Highly tasks/application-specific
• Generally not highly user-programmable
• Widely used in
• sensing and control
• real time operations
• Must be designed to work at low power in
  autonomous, distributed, networked and highly
  reliable manner.
• Impart limited "intelligent" characteristics to
  sensors
• adaptability to the changing physical world,
• real time, user-unsupervised functioning.
• Minimize number of parts and cost as compared to
  analog electronics
• Problem higher power requirements
• At design level, embedded systems require
• highly interactive hardware - software co-design,
  
• requires new embedded software tools that and
  testing systems that cannot be separated from
  physical aspects of a device application.
• (For example, embedded software controlling an
  infusion pump must be co-developed and tested
  with all other components - motors, actuators,
  sensors, fluid conduits, etc.)
• employ real time operating systems (RTOS) that
  must be highly fault tolerant
• In portable devices, are highly limited in
  available power.

4
Embedded processors in medical devicesExamples
of system developed or under development at Sarcos
5
Drug delivery devices with embedded controllers
6
Sensors and controllers 6 DOF Digital Load Cell
7
Imaging and hyperspectral systems Sarcos Micro
Camera
8
Strain, Multi-axis Strain and Rotary Sensors
Networks
RDTTM IC and Emitter
UASTTM IC Chip
Packaged Rotational Displacement Transducer
(RDTTM)
UniAxial Strain Transducer (UASTTM)
BiASTTM-based 6 DOF Force-Moment Sensor
BiASTTM IC Chip
BiAxial Strain Transducer (BiASTTM)
9
Telemedicine / telemonitoring systems sensing,
signal and data processing, communication,
interfaces
10
Medical embedded systemsFocus

• Functions
• Recovering and interpreting biological signals
• Optimal data sampling
• Communication major power drain
• Internal wired
• External wired or wireless networks
• Constraints
• Ultra-low power operation

11
Technical and scientific issues in embedded,
battery-powered systemsRecovering and
interpreting biological signals

• Effectiveness of current data recovery and
  interpretation algorithms in wearable,
  autonomously-operating medical devices that rely
  on sensor-body contact is limited by
• interfacial motion caused by the transducer
  motion (relative to skin and muscles), or passive
  and active skin/muscle motion that results in
  motion-related variations of the biological
  signal
• changes in the interface chemistry/electrochemistr
  y resulting in generation of electrical noise,
  and drift
• changes in the interface geometry resulting from
  tissue dimensional changes (e.g. from swelling or
  dehydration)
• changes in electrical characteristics of the
  interface caused by metabolic and physical
  factors (e.g. occlusion of the skin site by the
  transducer, changes in hydration, blood flow)
• frequently scarce "on-sensor" computational
  resources
• non-linear phenomena
• multidimensional data forming complex,
  time-dependent patterns, such as neuronal
  recordings in the CNS
• limited available power, bandwidth and physical
  space

12
Technical and scientific issues in embedded
battery-powered systemsRecovering and
interpreting biological signals

• Future capabilities of smart embedded sensors in
  battery-powered autonomous, wearable medical
  systems
• Optimal signal sampling to minimize power and
  storage needs
• Recovering (and interpreting) signals by
  combining inputs from multiple data sources in
  order to extract information.
• Performing data reduction, conditioning,
  interpretation and storage locally, adaptively,
  and "on-demand".
• enable distributed, adaptive sensor and control
  networks.
• Recognizing multidimensional patterns and
  interpreting information.
• Detecting, interpreting and minimizing motion
  artifacts,
• E.g. via sensor networks, (especially biomedical
  sensor networks), that maximize reliability of
  information derived from orthogonal sensors.
• Assessing quality and reliability of data
• Assessing confidence in measured data, especially
  in uncontrolled and noisy environment
• Intelligent power management.

13
Physiologic Data Acquisition and HR Recovery
Pulse rate recovery (lower trace) from ECG with
motion artifact-related noise (upper trace
Example of motion-induced interference
Pulse-pressure wave for cuff-free Blood Pressure
measurement ECG (ISU-based sensor, upper trace)
and pressure wave (finger-mounted sensor, lower
trace)
14
Sensor output, ECG and Breathing
15
Example Data interpretation algorithm
Extensive use of embedded processors
Injury assessment Normal Dead Injured Minor
trauma Major trauma Flags H/S PT/HT
Other Confidence measures
Severity of Injury Hemorrhagic Shock Other
conditions ALGORITHM
Trigger
on demand
HS hemorrhagic shock PT/HT pneumo/hemothorax
16
(No Transcript)
17
Technical and scientific issues in embedded
battery-powered, autonomous systemsSignal
Sampling

• Current methods
• Equidistant sampling
• Frequency-dependent sampling
• Problem
• Large number of data points
• Data storage requirements
• Available memory
• Speed
• Power consumption
• Data transmission
• Requires high bandwidth if sampling is not
  optimized
• Future Optimal, adaptive sampling algorithms
• Minimize power consumption, data storage and
  communication bandwidth requirements
• Enable high fidelity reconstruction of data in
  host computer
• Example
• ECG data acquisition, compression and
  reconstruction
• Image processing
• Neuronal recordings EMG, EEG.

18
Non-equidistant sampling(work in
progressco-author Dr.K.Sikorski, University of
Utah)
Equidistant sampling 10k points
Optimal sampling sampling 200 points
co
19
Technical and scientific issues in embedded
battery-powered systemsPower

• Currently
• Only few microprocessor families feature
  ultralow power consumption
• Requirement for battery powered, long lifetime,
  and remotely-operating (unsupervised) systems
  significantly limits
• Functionality of current embedded systems
• Processor selection and system design.
• Issues
• Safety and reliability at power limits
  potential for life-threatening situations
• Brown-outs/blackout recovery
• Assessment available power
• Loss of data
• Volume and weight
• Batteries are large
• Next generation of smart embedded sensors
  operating autonomously in wearable medical
  systems must
• Minimize power consumption of
• Sensors
• Data processing,
• Data communication
• Operate, with full-functionality, on scavenged,
  or biologically-generated power.
• Current power scavenging systems are inadequate
  (mechanical, thermoelectric, photovoltaic)

20
Example Power Consumption in devices with
wireless data transmission

• Integrated Sensor Unit
• ECG acq.
• waveform recovery
• Temp
• Body motion
• Body position
• Wireless comm

21
Example Power Consumption Comparison

• Blood pressure measurements
• 1. Oscillometric Method
• a. Cuff based (wrist size) 2.4 mAh per reading
  ( commercial product specification)
• Battery life approx. 400 readings
• b. Cuffless mode 0.2 mAh per reading
• Battery life approx. 4750 readings (approx. 2
  yrs)
• 2. Velocity measurement or pulse analysis (ToF)
  methods
• Less than 300 µA
• Battery life more 2 yrs expected
• SpO2 measurement Two-Wavelength Optical Method
• 1. Transmission mode 0.25 mAh per reading
• Battery life approx. 2 years
• 2. Reflectance mode 0.35 mAh per reading (1
  min. reading)
• Battery life approx. 1.5 years
• 3. Blood Pressure and SpO2 Combined Sensor
  Wrist Module
• Power consumption estimate 0.45 mAh per reading
  ( one minute)
• Battery life 1 year
• NOTE 1. Coin battery 950 mAh 2. Monitoring
  Modes
• Emergency cycle Continuous monitoring 1/ minute
  for 6 hours, 3 times repeat

22
Summary

• Current practical limitations of embedded,
  battery-powered medical systems
• Power requirements
• Limited availability of devices capable of
  ultra-low power operation
• Lack of biologically-derived power sources.
• Data acquisition and processing
• Lack of optimal sampling methods
• Signal recovery and interpretation algorithms for
• Multidimensional, non-linear signals.
• Future. Embedded processing will form basis for
• Adaptability to environment / task - Intelligent
  transducers
• Total integration of analog transducer
  functionality with signal conditioning, data
  processing, communication and interpretation.
• Multidimensional, real time pattern recognition
  and interpretation
• Learning.
• Plasticity of embedded processor networks
• Distributed, self-organizing, reconfigurable,
  adaptive, self-healing systems.
• Reliable and adaptive data collection, routing,
  storage and control.
• Systems capable of generating own adaptive,
  autonomous software in response to environment
  and/or task.

23
(No Transcript)
24
Bio

• Tomasz J.Petelenz, Ph.D., obtained his MS degree
  in physics microelectronics from the Silesian
  Technical University, Gliwice, Poland, and a
  Ph.D. degree in bioengineering from the
  University of Utah in Salt Lake City. He has
  over twenty years experience in medical and drug
  delivery device RD, working on projects ranging
  from iontophoretic transdermal drug delivery,
  kidney dialysis machines, infusion and injection
  devices, to non-invasive physiologic sensors,
  wireless data communication and prosthetics.
  Prior to joining Sarcos, he conducted
  iontophoretic research at the Center for
  Engineering Design, University of Utah, and was a
  Director of RD at Iomed, Inc. He is currently a
  Vice President of Medical Projects at Sarcos
  Research Corporation, managing biosensor and drug
  delivery development projects, and an Adjunct
  Associate Professor at the Department of
  Bioengineering, University of Utah . Dr.Petelenz
  is an inventor/co-inventor on 25 patents, and
  authored/co-authored 37 publications and
  presentations.