Transcript Slide 1

Measuring Ocular Microtremor
The Brainstem
• Has 3
components
• 4 cranial nerves
on each section
Assessing Brainstem Activity
• “High Tech”
 BIS Technology
 Expensive (sensors & trained technicians)
 Can’t use if swelling from head injuries
 Glasgow Coma Scale
• “Low Tech”
 Cranial Nerves
Unconscious Patients
Very uncomfortable procedures to
assess intactness of cranial nerves

Flush 60 oz. of ice cold water in ear
Ocular Microtremors (OMT)
 OMT have been shown in clinical studies to
accurately reflect brainstem activity
 Ocular motor nerves originate in the
brainstem where sensory signals are preprocessed and sent to the cerebrum
 Oculo-motor nerves originate in the
brainstem
 These signals are necessary for consciousness
Objective 1
Packaging &
Architecture
• Incorporate a
disposable element
into the sensor
mounting apparatus.
• Redesign and re-
package the
electronics for
miniaturization and
packaging for in-line
dongle-type element
on the cable.
Objective 2
Signal
Processing
- Signal processing for
signal amplification
and/or noise reduction
Objective 3
Integration Into
Patient Overhead
Monitor
• Research and understand the
standard(s) used to integrate into
this technology
• Implement in software with
potential hardware modifications
for the physical interface.
Current Trend Analysis
Software
Component Overview
Waveform
Generator
Test MUX
OMT
Simulator
Philips
VueLink
Module –
M1032A
Sensor
(Provided)
Analog Input/
Preprocessing
Circuitry Amp
For MCU, currently looking
at Atmel AVR32 with 12-bit
A/D and integrated DSP
functions,50 MHz clock.
Bedside Monitor
(Philips MP-60/70
or Agilent V24/26)
Power Supply
Microprocessor
with Integrated
A/D, UART, and
DSP
XRAM
OMT Sensor
 Composed of a piezoelectric transducer, and a surface
mount IC amplifier
 The sensor generates voltage when it undergoes stress,
in this case eye movement
 The piezoelectric transducer was designed to be really
sensitive to slight amounts of stress
Power Management
 The current “Blackbox” is powered by a medical grade
power supply which provides ±12V for the sensor, and
5V for the processing unit
 The Atmel AVR32 processor has a maximum voltage
rating of 3.6V
 The interface module to the bedside monitor will be
powered by the monitor itself
Power Management continued
 Originally the bedside monitor was going to be used to
power both the processor and sensor
 Power options
 Dual PS/2 – output voltage 5V ±10%
 Dual MIB/RS232 – output voltage 5V ±5%
 USB 2.0 – low power mode, output voltage 4.4V
 It is more convenient to construct our own custom
power supply that is capable of powering both the
sensor and processor
Power Flow
Sensor
Signal processing
unit
Monitor interface
Power supply
Bedside monitor
Broad Overview of
Software
Sample Signal until
XRAM Buffer Full
Perform DSP Algorithm
on Buffer Contents
Extract Necessary
Data From DSP
Output
Send Wave Samples,
Freq, and Amp. to
UART(Monitor)
Sampling
 Sampling will be interrupt
driven, and will continue until a
XRAM
buffer a buffer is filled. Buffer
must be large enough for an
adequate analysis of signal.
Buffer 1
 When the first buffer is full, the
A/D
Buffer 2
DSP
DSP
Output
Bus
DSP algorithm will be started on
the signal and sampling will
continue to a secondary buffer of
equal size. Sampling and DSP
will alternate between buffers so
no data is lost.
 The buffer size and sampling rate
will be configurable over the
UART.
FFT – Generate
Output Array in
XRAM
Search Valid
Range of Array
for Max Energy
FFT Method
Amplitude/
Frequency
Output
-The array locations corresponding
to the valid OMT range are searched
for the maximum energy, giving the
Frequency and Amplitude of the
OMT signal.
Zero Out Locations
Outside of OMT
Range (Filter)
Inverse FFT to
Obtain Filtered
Waveform in
Time Domain
-An FFT is performed on the
contents of the sample buffer, with
spectrum data saved to an array.
Filtered
Waveform
Output
-An inverse FFT on only the valid OMT
frequency range is required to
reconstitute a time-domain waveform
that can be sent to the bedside
monitor.
Peak-Detection Method
Digital
HPF/LPF/Notch
Filter Routines
Modify Buffer
Peak-Count and
AmplitudeAveraging
Algorithm
Filtered
Waveform
Output
Amplitude/
Frequency
Output
Wavelets – Short Oscillations of a given
frequency. Can be compared to our signal
to determine if energy at that frequency
is present.
-Digital filter routines modify the
contents of the buffer, giving a filtered
signal that could be sent to the bedside
monitor.
-Positive-to-negative changes in slope
are counted, giving the frequency. The
amplitudes at which this occurs are
averaged.
Bedside Monitor Interface
 Philips VueLink Module 1032A
 RS-232 Input – Open Interface
Protocol
 Allows 1 Waveform (OMT
Signal), 2 Numerics
(Amplitude and Frequency),
Diagnostic Messages (ex.
OMT out of valid range), and
Alarms (ex. OMT below
certain value)
http://www.sentec.ch/fileadmin/documents/manuals/EN-HB-005928-bSDM_VueLink_Installation_Manual.pdf
Testing and Data Mining
 Important in order to determine the pattern and/or
trend of OMT signals
 Determining a pattern is vital to the signal processing
component
 The process involves a subject (one of us) to lie down
and have the sensor mounted on a closed eyelidusing
surgical sensitive tape
 Due to a lack of FDA approval and other required
certification, we will not be able to gather data on a
large scale basis
Vivian’s OMT
Mark’s OMT
Division of Labor
Mark
Power
Power Supply
Powering Sensor
MCU
Signal Sampling - A/D
DSP of Signal
Interface
Analog Sensor
Output Interface (monitor)
Manufacturing
PCB
Dongle
Administrative
Mediary to EyeTect
Mediary to medical experts
Documenatation
Technical Manual
User's Manual
UROP/EEF
Extra (if time)
Data Mining/Trend Evaluation
Bedside Monitor
Tom
x
x
x
x
x
x
x
x
Vivian
Salman
Steve
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Schedule
First Milestone
Grants
Second Milestone
Pre-Expo
Throughout Semester
Budget
Expenses
Product
Unit Cost Amount Total Cost
Provider
Processing/Manufacturing
Atmel AVR32 Development Kit
2-layer PCB
Packaging
179.00
33.00
20.00
1
3
1
179.00
99.00
20.00
Vuelink M1032A
Philips M3 Bedside Monitor
300.00
2,650.00
1
1
Capacitors
Diodes
Power MOSFET Transistor
Cooling Fan/Heat Sink
Power MCU
Packaging
4.25
42.00
74.00
20.00
10
10
10
1
1
1
30.00
30.00
42.50
42.00
74.00
20.00
Printing Manuals
Poster
Shipping/Handling Fees
40.00
75.00
200.00
1
1
1
40.00
75.00
200.00
http://mouser.com/
http://www.4pcb.com
Interface
300.00 http://www.hp.com
2,650.00 http://www.dotmed.com
Medical Grade Power Supply
Administrative Costs
Margin
7%
266.11
Total
4,067.61
Funding
Outside Funds
EyeTect
UROP
3,000.00
?
Interference Concerns
 Accidental patient movement
 Electrical noise from other equipment in the ICU
room
 Noises in excess of 95 dB
 Heartbeat
Contingencies
 Software implementation of display if bedside
monitor is not able to be acquired
 Scaling number of processors to ensure proper
operation
 Purchasing additional components to ensure
minimal downtime especially before expo
Risks
 During application of sensor patients’ eye may
be agitated
 Additional cord is hanging off of patient
 Possibility of dongle removing sensor from
patients’ eye if too heavy
 Components may be destroyed if wired
incorrectly
Extension possibilities
 Add through connections in order to
minimize hanging cables
 Adding accelerometer to measure patient
movement to find bad samples
Questions?