Comprehensive Ultrasound Research Platform

1
Comprehensive Ultrasound Research Platform

• Emma Muir
• Sam Muir
• Jacob Sandlund
• David Smith
• Advisor Dr. José Sánchez

2
Outline

• Introduction
• Block Diagram
• Proposed System
• Functional Description
• Requirements
• Preliminary Work
• Equipment
• Schedule

2
3
Outline

• Introduction
• Block Diagram
• Proposed System
• Functional Description
• Requirements
• Preliminary Work
• Equipment
• Schedule

3
4
Ultrasound Introduction

• Medical Applications
• Detecting tumors and abnormalities
• Piezoelectric Transducer
• Pulse Excitation
• Changes in density reflect waves

4
5
Outline

• Introduction
• Block Diagram
• Proposed System
• Functional Description
• Requirements
• Preliminary Work
• Equipment
• Schedule

5
6
Block Diagram
6
7
Outline

• Introduction
• Block Diagram
• Proposed System
• Functional Description
• Requirements
• Preliminary Work
• Equipment
• Schedule

7
8
System Requirements

• Up to 8 transducer channels
• Excitation waveforms 3 µs or less
• Time-bandwidth product of 40
• Design for high frequency
• Signal to noise ratio (SNR) gt 50 dB

8
9
Block Diagram
9
10
Waveform Generation

• Resolution Enhanced Compression Technique (REC)
• Pre-enhanced chirp calculated with convolution
  equivalence
• Increase Bandwidth (BW) of outputted signal
• Improves resolution

10
11
REC Technique

• h1(n) c1(n) h2(n) c2(n)
• h1(n) Transducer Impulse Response
• h2(n) IR with increased BW
• c2(n) Linear chirp
• c1(n) Calculated pre-enhanced chirp

11
12
REC Requirement

• Enhance the bandwidth of the transducer to 12.45
  MHz
• Original bandwidth 8.3 MHz
• Increase resolution

12
13
Block Diagram
13
14
Sigma Delta Modulation

• Analog to digital conversion technique
• 1-bit ADC
• Oversampling
• Quantization error compensation
• Signal values either 1 or -1
• Sigma Delta Toolbox

15
Sigma Delta Modulation
16
Sigma Delta Modulation
17
Sigma Delta Modulation

• 10 mean squared error
• Chirp signal from 4MHz to 12MHz
• Based on REC signal
• 1.024 Gsamples/second
• FPGA sampling rate 1.06 Gsamples/second
• Oversampling Rate (OSR) 512
• T 1 µs for testing, 3 µs for REC signal
• OSR must be a power of 2
• 1.024 GSamples/second 2OSR/T
• Divide Amplitude of signal by 2
• Avoid overloading
• High OSR
• High order

18
Sigma Delta Modulation
19
Sigma Delta Modulation
20
Block Diagram
20
21
Why FPGA?

• Array, high speed
• 8-pins
• gt 600 MHz
• Accurate, uninterrupted transmission
• Flexibility

22
FPGA Description

• Store waveform data
• Needs to be high speed
• Individual data for each pin
• Parallelize to pins
• High-speed transmission
• 1.07 GHz

22
23
FPGA Requirements

• Connect to PC
• 24 kbits
• Store on DDR2
• 62.5 MHz required (266 MHz actual)
• At least 8 waveforms
• 3000 bits per waveform
• Each pin individualized (8 pins)
• Up to 5 ms delays
• Different waveforms

23
24
FPGA (Pseudo) Flow Chart
24
25
FPGA Block Diagram
25
26
FPGA System
26
27
FPGA Transmission
27
28
FPGA to PC Communication

• UART
• 115200 baud
• Send waveform data
• Assign waveform to pins
• Assign delay to pins
• Start transmission

28
29
Block Diagram
29
29
30
High Voltage Amplifier

• Two parts
• Operational Amplifier
• H-bridge
• Op Amp
• Amplify the signal from the FPGA
• -1 V to 1 V, or 0 V to 3.3 V
• Amplify to 10 V
• Needed for H-bridge
• Slew Rate

30
31
High Voltage Amplifier

• H-bridge
• MOSFET configuration
• Will amplify from 10 V to 100 V
• 100 V is a safe threshold for Ultrasound
• Must work at 1 Gsample/sec
• Output to ultrasonic transducer (or LPF)

31
32
Block Diagram
32
32
33
Low Pass Filter

• Convert the sigma delta signal to analog
• RC circuit will be used
• Bandpass nature of the source could be the filter
  
• More research needed

33
33
34
Block Diagram
34
34
35
Ultrasonic Transducer

• 128-channel linear array
• Will only be using 8 channels

35
36
Block Diagram
36
36
37
T/R Switch

• Protects analog front end from high voltages
• Clamps voltages so damage is avoided
• 1.9 Vpp is specified
• TX810 by Texas Instruments

37
38
Printed Circuit Board

• Design
• Favor over perforated and bread board
• Six layers predicted because of frequencies
• Eliminates cross-talk and EMI
• More research is being done to predict layout of
  the board

38
38
39
Block Diagram
39
39
40
Analog Front End

• Two Parts
• Low Noise Amplifier (LNA)
• Analog-Digital Converter
• Using AD9276-80KITZ provided by Analog Devices
• Contains both parts
• LNA
• Will amplify the received signal from the
  ultrasonic transducer
• Low amplitude from the transducer
• High SNR

40
40
41
Analog Front End

• A-D
• Convert analog signal
• Will output digital signal to embedded device

41
42
Block Diagram
42
43
PC Data Processing
43
44
PC Data Processing
44
45
Pulse Compression

• Improve penetration depth and SNR
• Techniques
• Matched filter
• Optimal for large amount of noise
• Cross correlation
• Creates side-lobes
• Inverse filter
• Optimal for zero noise
• Inaccurate for large amount of noise

45
46
Pulse Compression

• Wiener filter
• Balance between matched and inverse filters based
  on SNR
• Smoothing parameter
• Can be used to optimize filter
• Adjusts weighting of matched and inverse filter
  components

46
47
PC Data Processing
47
48
Beamforming
Sensor 1
Sensor 2
Focus Distance (FD)
Sensor 3
Sensor 4
Sensor 5
Sensor 6
Sensor 7
Sensor Distance (SD)
Sensor 8
48
49
Beamforming

• Delays based on Focus Distance (FD) and Sensor
  Distance (SD)
• ?t1 (FD2 (½SD)2)0.5 FD(2/1540)
• ?t2 (FD2 (1½SD)2)0.5 FD(2/1540)
• ?t3 (FD2 (2½SD)2)0.5 FD(2/1540)
• ?t4 (FD2 (3½SD)2)0.5 FD(2/1540)
• 1540 m/s is the speed of sound in tissue
• Multiply by 2 to account for distance travelled
  in both directions

49
50
Beamforming
Sensor 1
Sum
Sensor 8
Sensor 2
Sum
Delay ?t4 -?t3
Sensor 7
Sum
Delay ?t4 -?t2
Sensor 3
Sum
Output
Sensor 6
Sensor 4
Sum
Delay ?t4 -?t1
Sensor 5
50
51
PC Data Processing
51
52
Time-Gain Compensation

• Compensates for attenuation of the received
  signal
• Attenuation of the sound waves is caused by the
  depth of the echoing substance
• More depth More attenuation

52
53
PC Data Processing
53
54
Envelope Detection

• Determines the bounds of the processed signal
• Detected width contains the display information
  about the tested tissue
• Hilbert transform

54
55
PC Data Processing
55
56
Log Compression

• Convert the data from linear values to dB values
• 20log10(Current Value)
• Creates clearer images

56
57
PC Data Processing
57
58
Graphical User Interface (GUI)

• Displays image result of signal processing
• Allows user to enter contrast
• Allows user to select depth

58
59
Graphical User Interface (GUI)
Contrast
Current Displayed Image
Max

Update


0.25cm
10cm
20cm
30cm
Depth
59
60
Graphical User Interface (GUI)

• Functional Requirements
• All data processing shall be performed in less
  than 2 minutes.
• The image created will display an image for
  depths between 0.25 cm and 30 cm.

60
61
Outline

• Introduction
• Block Diagram
• Proposed System
• Functional Description
• Requirements
• Preliminary Work
• Equipment
• Schedule

61
62
Block Diagram
62
63
Sigma Delta Modulation

• 44th Order Equiripple Filter without Gain
  Compensation

64
Sigma Delta Modulation

• 44th Order Equiripple Filter with Linear Gain
  Compensation

65
Sigma Delta Modulation

• RC (R330 O, C10 pF) Filter without Gain
  Compensation

66
Sigma Delta Modulation

• RC (R330O, C10pF) Filter with Linear Gain
  Compensation

67
Sigma Delta Modulation

• Cross Correlation and Mean Square Error

Filter Cross Correlation Mean Square Error
Equiripple filter without Gain Compensation 0.9888 -
Equiripple filter with Linear Gain Compensation 0.9900 -
RC without Gain Compensation 0.9129 0.0264
RC with Linear Gain Compensation 0.8418 0.0387
68
Block Diagram
68
69
PC Data Processing
69
70
PC Data Processing

• Array of points in Fields II

70
71
Beamforming
71
72
Time-Gain Compensation
72
73
Envelope Detection
73
74
Log Compression
74
75
Log Compression
75
76
Graphical User Interface (GUI)
0
10
20
30
40
50
Distance in mm
60
70
80
90
100
76
77
Graphical User Interface (GUI)
Contrast
Max

Update


0.25cm
10cm
20cm
30cm
Depth
77
78
Block Diagram
78
79
REC Preliminary Results

• MATLAB simulation
• Increased bandwidth of transducer to 150 of
  original bandwidth
• Linear chirp frequencies in the range of 1.14
  times the bandwidth
• Optimal number to reduce side-lobes during pulse
  compression
• Resulting chirp can be applied to finished system

79
80

• h1(n) c1(n) h2(n) c2(n)

80
81
81
82
Block Diagram
82
83
PC Data Processing
83
84
Pulse Compression Results

• MATLAB simulation
• Wiener filter
• SNR of 60 dB
• Input is REC pre-enhanced chirp

84
85
85
86
Block Diagram
86
87
FPGA Preliminary Work

• Interface to DDR2
• System to arbitrate access
• Multi-pin high-speed output
• Verified at lower frequencies
• Separate data, delays for pins
• UART works alone, needs integration

87
88
Block Diagram
88
88
89
Analog Preliminary Work

• H-bridge
• VDD 15 V VDS 10 V VGS 5 V ? Vd 5 V
• Datasheet at those values Id ? 0.3 A, ?
  Rd ? 20 ?.
• PRd (0.3 A)2 x (16.667 W) 1.5 W.
• High for resistors used

89
90
Analog Preliminary Work

• MOSFET circuit used

90
91
Analog Preliminary Work

• H-bridge device cannot function at high enough
  frequencies
• Decided to build own H-bridge

91
92
Block Diagram
92
92
93
Analog Preliminary Work

• Simulation circuit for T/R Switch
• Spice model provided by T.I.

93
94
Simulated waveforms
1.9V?
Simulated at Vin 10V. Voutpp 1.85V
94
95
Simulated waveforms
1.9V?
Simulated at Vin 90V. Voutpp 1.94V Includes
overshoot.
95
96
Printed Circuit Board

• Design changes (i.e. H-bridge), so board changes
• Still no connector for the transducer
• Using OrCAD PCB Designer
• Book provided to assist with design
• IC footprint troubles

96
96
97
Equipment

• LeCroy High Speed Oscilloscope 725Zi
• Blatek 128 pin Ultrasound Probe
• Ultrasound Testing Phantom
• Xilinx Virtex 5 Development Kit ML509
• UART Null Modem Adapter
• Analog Devices Analog Front End AD9276-80KITZ
• Low noise amplifier
• Variable control amplifier
• ADC
• MOSFETS x4 for the H-Bridge model to be
  determined
• PCB x2
• Texas Instruments T/R Switch TX810
• Software
• Matlab Version 7.5.0.342 R2007b
• Sigma Delta Toolbox
• Field II
• Agilent Connection Expert
• Xilinx
• Hydrophone

97
98
Schedule
98
99
Acknowledgments

• The authors would like to thank Analog Devices
  and Texas instruments for their donation of
  parts.
• This work is partially supported by a grant from
  Bradley University (13 26 154 REC)
• Dr. Irwin
• Dr. Lu
• Mr. Mattus
• Mr. Schmidt

99
100
References

• 1 J. A. Zagzebski, Essentials of Ultrasound
  Physics, St. Louis, MO Mosby, 1996.
• 2 R. Schreier and G. C. Temes. Understanding
  Delta-Sigma Data Converters, John
• Wiley Sons, Inc., 2005.
• 3 R. Schreier, The Delta-Sigma Toolbox Version
  7.3. Analog Devices, Inc, 2009.
• 4 T. Misaridis and J. A. Jensen. Use of
  Modulated Excitation Signals in
• Medical Ultrasound, IEEE Trans. Ultrason.,
  Ferroelectr. Freq. Contr., vol. 52, no. 2,
• pp. 177-191, Feb. 2005.
• 5 M. Oelze. Bandwidth and Resolution
  Enhancement
• Through Pulse Compression, IEEE Trans.
  Ultrason., Ferroelectr. Freq. Contr., vol. 54,
• no. 4, pp. 768-781, Apr. 2007.
• 6 Mitzner, Kraig. Complete PCB Design Using
  OrCad Capture and PCB Editor,
• Newnes, 2009.

100
101
References Cont.

• 7 Montrose, Mark I. Printed Circuit Board
  Design Techniques For EMC Compliance
• A Handbook for Designers, Wiley-IEEE Press, 2000.
• 8 J.A. Jensen Field A Program for Simulating
  Ultrasound Systems, Paper presented
• at the 10th Nordic-Baltic Conference on
  Biomedical Imaging Published in Medical
• Biological Engineering Computing, pp. 351-353,
  Volume 34, Supplement 1, Part 1,
• 1996.
• 9 J.A. Jensen and N. B. Svendsen Calculation
  of pressure fields from arbitrarily
• shaped, apodized, and excited ultrasound
  transducers, IEEE Trans. Ultrason.,
• Ferroelec., Freq. Contr., 39, pp. 262-267, 1992.

101
102
Questions?
102