1. Ultrasonic Nonlinear Imaging- Tissue Harmonic Imaging

2. Conventional B-mode image (AP4CH)

3. THI
Fundamental

4. THI
Fundamental

5. Tissue Harmonic Imaging

6. Tissue Nonlinear Imaging
Performance of ultrasound has been sub-optimal on technically difficult bodies.
Most recent new developments have bigger impact on technically satisfactory bodies.
Poor image quality leads to uncertainty in diagnosis and costly repeat examinations.
Tissue Harmonic Imaging

7. Tissue Harmonic Imaging
Methods to improve image quality:
Different acoustic window.
Lower frequency.
Adaptive imaging.
Non-linear imaging (or harmonic imaging).
Tissue Harmonic Imaging

8. Origin of Tissue Non-linearity
Finite amplitude distortion: peaks of the waveform travels faster than the troughs.
Tissue Harmonic Imaging

9. Tissue Non-Linearity Signal Source
Finite amplitude distortion generated tissue harmonics t
Pressure
Before distortion
After distortion

10. Non-Linear Propagation
Tissue Harmonic Imaging

11. Tissue Harmonic Imaging
Axial Amplitude 20 40 60 80
Depth(mm)
Velocity(cm/sec)
Fundamental
2nd Harmonic
Tissue Harmonic Imaging

12. Tissue Non-Linearity THI Characteristics 4MHz Beam Patterns transducer
Lateral Position (mm)
4MHz Beam Patterns dB

13. Tissue Harmonic Imaging
Tissue Non-linearity
Tissue harmonics are virtually zero at the probe face.The intensity continues to increase until attenuation dominates.
The higher the intensity is, the more tissue harmonics are generated.
Such a mechanism automatically increase the difference between signal and acoustic noise.
Tissue Harmonic Imaging

14. Advantages of Tissue Harmonic Imaging
Low sidelobes.
Better spatial resolution compared to fundamental imaging at the original frequency.
Less affected by tissue inhomogeneities – better performance on technically difficult bodies.
Tissue Harmonic Imaging

15. Non-linear Parameter B/A
B/A defines non-linearity of the medium. The larger the B/A, the higher the non-linear response.
Tissue Harmonic Imaging

16. B/A Parameters: Measurements
Finite amplitude method:
B/A is related to the second harmonic generation. Thus, it can be found by relating the signal amplitude at the fundamental frequency to the second harmonic component.
Thermodynamic method:
The B/A value is determined by measuring the change of sound speed with pressure and temperature.
Tissue Harmonic Imaging

17. B/A Parameters: Typical Values
Water:5.5+/-0.3.
Liver: 7.23.
Fat: 10.9.
Muscle: 7.5.
Results from both methods have excellent agreement.
B/A imaging may be used for tissue characterization.
Tissue Harmonic Imaging

18. Tissue Harmonic Imaging
Image Analysis Issues
Low signal-to-noise ratio: coded excitation, simultaneous multiple transmit focusing.
Spectral leakage and image quality degradation.
Spatial covariance analysis for correlation-based processing.
Motion artifacts in pulse inversion imaging .
Tissue Harmonic Imaging

19. Filter Based Image Formation
Fundamental and Harmonic Imaging
Fundamental Imaging
Spectrum
Spectrum
MHz
MHz
LPF
Harmonic Imaging
Transmit Signal
Received Signal
HPF

20. Effects of Harmonic Leakage
Motive :
Contrast resolution degradation
4MHz Beam Patterns
MHz
MHz dB
MHz
MHz
Transmit Spectrum
Received Spectrum
Lateral Position (mm)

21. Sources of Harmonic Leakage
Designed transmit waveform.
System nonlinearity.
Electromechanical conversion.
Waveform Generator
High Voltage Amplifier
&
T/R Switch
Transducer
There are several sources of harmonic leakage. One is the inherent leakage contained in the designed transmit waveform. In other words, the original transmit waveform may already have frequency components in the second harmonic band. The second source of harmonic leakage is due to the nonlinearities of the imaging system. For example, the high voltage amplifier and transmit/receive switching may distort the waveform and produce harmonics. The third source is the nonliear response of the transducer.
Tissue Harmonic Imaging

22. Tissue Harmonic Imaging
Designed Waveform (I)
Characteristics of transmit waveforms.
Waveforms
Normalized
Amplitude s
We will first investigate effects of harmonic leakage due to the characteristics of designed transmit waveforms. Showing in the upper panel of this slide are three different waveforms. All waveforms have a center frequency of 2MHz. The yellow curve has Gaussian envelope with 25% fractional bandwidth, the purple curve is a gated sine wave with 4 cycles and the green curve is a gated square wave also with 4 cycles. The corresponding spectra are shown in the lower panel. As you can see, all three waveforms have similar bandwidth around the center frequency and the Gaussian waveform has negligible frequency component in the harmonic band. The other two waveforms, however, have significant harmonic leakage signals primarily due to the nature of the envelope.
Spectra dB
MHz
Tissue Harmonic Imaging

23. Designed Waveform (II)
Signal bandwidth.
MHz
Spectrum of Transmit Signal
Fundamental
band
Harmonic
Leakage
The other possibility to have harmonic leakage is the wide signal bandwidth. Even for Gaussian signals, the fundamental band may overlap with the harmonic band if the bandwidth is sufficiently large. Such leakage signal may also be mixed with the desired tissue harmonic signal and potentially affect characteristics of the radiation pattern.
Tissue Harmonic Imaging

24. Non-linear Propagation
Wave at distance z
angular spectrum method
Linear propagation to z+Dz
frequency domain solution to Burgers’ equation
Nonlinear propagation at z+Dz
Tissue Harmonic Imaging

25. Nonlinear Simulation Model
Model the Nonlinear Propagation
Δf: fundamental frequency
un： Sin(2π(nΔf)t)
β：nonlinear parameter
c：sound velocity

26. Results: Effect of Bandwidth
Gaussian at 25% and 50%
Contrast v.s. Spatial
Harmonic Beam Patterns
BW=25%
BW=50%
-20 dB
-40
-60
-10 -5 5 10
Lateral Position (mm)

27. Results: Signal Type Sine, square and Gaussian wave, BW=25%
Smooth envelope has better contrast
Harmonic Beam Patterns
Gaussian
Gated sine
Gated square
-20 dB
-40
-60
-10 -5 5 10
Lateral Position (mm)

28. Integrated Harmonic Beam Patterns
Results: Signal Type
Gaussian, gated sine and gated square waves.
BW=50%.
Harmonic Beam Patterns
Integrated Harmonic Beam Patterns dB
Gaussian
Gated sine
Gated square
4MHz linear
This slides shows similar comparisons as the previous slide except that the bandwidth is increased from 25% to 50%. There are four radiation patterns, (yellow) Gaussian, (green) gated square, (purple) gate sine and the red curve is the radiation pattern corresponding to a linear beam at the second harmonic frequency. We find that at a larger bandwidth, the differences in sidelobes between the three tissue harmonic beam patterns become smaller. Nevertheless, all the three patterns still have lower sidelobes than those of a linear beam of a Gaussian waveform at 4MHz. In other words, despite the harmonic leakage, the finite amplitude distortion based imaging may still outperform linear imaging with similar bandwidth. The sidelobe differences are also demonstrated in the figure on the right. Such a presentation is adopted from Dr. Ted Christopher. The horizontal axis is lateral position and the vertical axis is the normalized integration of the beam pattern intensity along the lateral direction. In other words, a curve rises rapidly to unity represents lower average sidelobes. From this curve, it is also evident that the Gaussian signal has the lowest sidelobes and all three nonlinear beams are better than the linear beam at 4MHz.
Lateral Position (mm)
Lateral Position (mm)
Tissue Harmonic Imaging

29. Effects of Harmonic Leakage Phase Aberration Pattern
Tissue Inhomogeneities
Fat layer: 15mm thick, B/A=10.
Aberrating plane: max. time delay error=30ns, correlation length=5mm.
Phase Aberration Pattern 30
-30
(ns) -5
-10 5 10
(mm)
2.5
-2.5

30. Results: Tissue Inhomogeneities
BW=50%.
Harmonic Beam Patterns
Integrated Harmonic Beam Patterns dB
Gaussian
Gated sine
Gated square
4MHz linear
This slides shows similar comparisons at a different bandwidth, which is 50%. Again, the red curve represents a 4MHz linear radiation pattern. At a higher bandwidth and with tissue inhomogeneities, the differences between the three waveforms are greatly reduced. On the other hand, all three waveforms outperform the linear case, as shown in both the radiation pattern and normalized integrated radial patterns.
Lateral Position (mm)
Lateral Position (mm)
Tissue Harmonic Imaging

31. Effects of Harmonic Leakage
Drive Voltage
Magnitude => nonlinearity 
Spectra
Beam Patterns
1 Volt
5 Volt
Harmonic(1 Volt)
Fundamental(1 Volt)
Harmonic(5 Volt)
Fundamental(5 Volt) dB dB
Frequency(MHz)
Lateral Position(mm)

32. Tissue Harmonic Imaging
Results: Bandwidth
Gaussian envelope, 1 Volt, 25% vs. 50%.
Spectra
Harmonic Beam Patterns
Lateral Position(mm) dB
BW=25%
BW=50%
BW=25%
BW=50% dB
This slide compares measurements corresponding to two Gaussian signals with two different bandwidth. Both have 1V amplitude before the signals are power amplified. The yellow curve has 25% bandwidth and the purple curve has 50% bandwidth. The spectra shown on the left were measured at the probe face and the second harmonic radiations patterns shown on the right were measured at the focal point. Again, a larger bandwidth produced higher sidelobes.
Frequency(MHz)
Tissue Harmonic Imaging

33. Results: Drive Voltage
1Volt vs. 5Volt.
Beam Patterns
Spectra
Harmonic(1 Volt)
Fundamental(1 Volt)
Harmonic(5 Volt)
Fundamental(5 Volt)
1 Volt
5 Volt dB dB
The first experiment was done to demonstrate potential effects due to system nonlinearity. We compared two different amplitudes, one is 1V and the other is 5V at the output of the function generator before being power amplified. Other than the amplitude, the two signals are both Gaussian with 25% bandwidth. The spectra measured by the hydrophone at the transducer face are shown in the figure on the left, with the yellow corresponding to the 1V signal and the purple indicating the 5V signal. Since there are measured at the probe face, the harmonic components represent the harmonic leakage that we have discussed. We expected that a high amplitude is more likely to be affected by system nonlinearity. This is consistent with the spectra, in which the signals in the harmonic band are slightly higher with the purple curve than the yellow curve. The measured radiation patterns are shown in the figure on the right. All patterns are normalized to the respective maximum The green and red curves, which are almost identical, are the patterns at the fundamental frequencies. On the other hand, the 5V harmonic pattern, represented by the purple curve is consistently higher than the 1V case. The differences between the two harmonic patterns are small, however, potentially due to the differences are mainly below –40dB level in the spectra.
Frequency(MHz)
Lateral Position(mm)
Tissue Harmonic Imaging

34. Tissue Harmonic Imaging
Pulse Inversion
Positive driving pulse
Fundamental signal t f
Harmonic signal
Nonlinear propagation t f
Negative driving pulse t f
Tissue Harmonic Imaging
ONLY harmonic signal

35. Pulse Inversion Pulse inversion reduces sidelobe levels
Fundamental Beam
Harmonic Beam (Filtering)
Harmonic Beam (Filtering)
Harmonic Beam (Pulse inversion)
Gaussian pulse
Gaussian pulse
Tissue Harmonic Imaging
Sine pulse
Sine pulse

36. Tissue Harmonic Imaging
Pulse Inversion
harmonic leakage could be avoided
all linearly propagated components are cancelled
Harmonic Beam Patterns at 50% Bandwidth dB
Tissue Harmonic Imaging
Lateral Position(mm)

37. Tissue Harmonic Imaging
Harmonic Leakage
Smooth envelopes provide lower sidelobes, but also require a more sophisticated transmitter.
Large bandwidths improve axial resolution, but also increase sidelobes.
Sidelobe differences decrease in the presence of tissue inhomogeneities.
Spectral leakage must be suppressed without affecting fundamental beams.
Pulse inversion technique is the most effective.
In conclusion, we have discussed effects of harmonic leakage on the radiation patterns in various conditions. First, smooth envelopes, such as the Gaussian envelope, provide lower sidelobes due to lower harmonic leakage. However, a relatively more complicated transmitter is also required compared to the generation of simple gated waveforms. In addition, we have demonstrated potential tradeoff between axial resolution and contrast resolution in tissue harmonic imaging, due to the increased sidelobes associated with larger signal bandwidth. In the presence of tissue inhomogeneities, sidelobe differences between different waveforms become smaller. In any case, the nonlinear beams outperform the linear beam at the same frequency. Finally, in order to further improve performance of tissue harmonic imaging, means of suppressing harmonic leakage may be of great values. However, the harmonic leakage must be suppressed without significantly affecting the fundamental beam. Since quality of the nonlinearly generated second harmonic beam is directly determined by the quality of the fundamental beam. Thank you.
Tissue Harmonic Imaging

38. Sound Velocity Inhomogeneities
v1 v2
Array Transducer
Tissue Harmonic Imaging

39. Spatial Covariance Analysis
Sound velocity inhomogeneities are conventionally corrected by correlation-based methods.
The covariance of signals received at different positions is critical to correlation-based correction techniques (the van Cittert-Zernike theorem).
Is it possible to further improve the image by combining tissue harmonic imaging and phase aberration correction?
Optimal frequency selection for imaging and time delay estimation.
Tissue Harmonic Imaging

40. Progress: Simulations
Transmit
beam formation by FDSBE
Receive
time-domain signal for each channel
Tissue Harmonic Imaging
a,b: length and width of channel

41. Progress: Simulations
time
channel
Correlation coefficient
channel
Tissue Harmonic Imaging

42. Progress: Results
Harmonic covariance is generally similar to or lower than fundamental covariance
0.5
0.2
0.4
0.6
0.8 1
Spatial Covariance: Simulations
Correlation Coefficient
2MHz Fundamental
4MHz Second Harmonic 1
0.5
0.2
0.4
0.6
0.8
Spatial Covariance: Experiments
3.5MHz Fundamental
7MHz Second Harmonic
Tissue Harmonic Imaging
Normalized Distance

43. Progress: Results With sound velocity inhomogeneities
Normalized Distance
0.5 1
0.2
0.4
0.6
0.8
Spatial Covariance : Experiments
3.5MHz Fundamental
7MHz Harmonic
Spatial Covariance: Simulations
Correlation Coefficient
2MHz Fundamental
4MHz Second Harmonic
Tissue Harmonic Imaging

44. Progress: Results Effects of SNR Correlation Coefficient
0.5 1
0.2
0.4
0.6
0.8
Fundamental Spatial Covariance: Experiments
Correlation Coefficient
Normalized Distance
High SNR
Low SNR
Second Harmonic Spatial Covariance: Experiments
Tissue Harmonic Imaging

45. Tissue Harmonic Imaging
Spatial Covariance
When adequate SNR is available
Whether the sound velocity variations are present or not, the harmonic covariance is generally similar to or lower than fundamental covariance.
When SNR is low
Harmonic covariance is significantly affected.
Imaging at the second harmonic frequency, correlation-based correction at the fundamental frequency.
Tissue Harmonic Imaging

46. Harmonic Interference
In contrast imaging, in which the tissue harmonic signals are un-desirable, the amplitude of the propagating wave needs to minimized.
Large apertures (smaller f-numbers) may be used.
It was reported that tissue harmonic signal can be reduced by 3dB by doubling the aperture size.
Tissue Harmonic Imaging

47. Harmonic Interference
Harmonic cancellation system: non-linear propagation is reduced by using a new signal at the harmonic frequency.
Phase and magnitude of the signal may be pre-calculated, but on-line adjustment is necessary.
Due to attenuation, optimal effects may only be achieved locally.
Tissue Harmonic Imaging

48. Ultrasonic Nonlinear Imaging- Contrast Harmonic Imaging
Tissue Harmonic Imaging

49. Contrast Harmonic Imaging
Contrast agents are used to provide higher contrast. The three commonly seen contrast agents are backscatter, attenuation and sound velocity.
Contrast agents could be solid particles, emulsion, gas bubbles, encapsulated gas, or liquid.
Tissue Harmonic Imaging

50. Contrast Harmonic Imaging
Primary clinical benefits:
Enhanced contrast resolution between normal and diseased tissues.
Outline of vessels or heart chambers.
Tissue characterization by using tissue specific agents.
Increasing blood flow signals.
Dynamic study using washout curve.
Tissue Harmonic Imaging

51. Tissue Harmonic Imaging
Example
Tissue Harmonic Imaging

52. Clinical Applications: Cardiology
Endocardial border detection.
Left ventricle (LV) function.
Valvular regurgitation quantification.
LV flow patterns.
Perfusion area of coronary artery.
Assessment of surgery for ventricular septal defect.
Tissue Harmonic Imaging

53. Clinical Applications: Others
Liver tumor enhancement.
Uro-dynamics and kidney functions.
Tubal function and placenta perfusion.
Transcranial Doppler enhancement.
LV pressure measurements.
Tissue Harmonic Imaging

54. Current Contrast Agents
Echovist.
Albunex.
Levovist.
Echogen.
Quantison.
Many more,…
Tissue Harmonic Imaging

55. Tissue Harmonic Imaging
Contrast Mechanisms
Strong backscattering produced by air bubbles.
The backscatter increases roughly linearly with the number of micro-bubbles.
A bubble in liquid acts as a harmonic oscillator. Acoustic resonance provides the major echo enhancement. In addition, strong harmonics are produced.
Tissue Harmonic Imaging

56. Tissue Harmonic Imaging
Contrast Mechanisms
Acoustic attenuation of soft tissues is typically represented by a constant (e.g., 0.5dB/cm/MHz).
Since contrast agents significantly change the scattering properties, attenuation measurements can also be used for contrast enhancement.
Tissue Harmonic Imaging

57. Tissue Harmonic Imaging
Contrast Mechanisms
Sound velocity is primarily determined by density and compressibility. Apparently, micro-bubble based contrast agents alter sound velocity.
Contrast enhancement based on sound velocity variations is still academic.
Tissue Harmonic Imaging

58. Tissue Harmonic Imaging
Contrast Mechanisms
Micro-bubbles produce strong harmonics when insonified near the resonance frequency.
If such harmonics are stronger than tissue harmonics, contrast can be improved.
Second harmonic signal is most useful due to limited transducer and system bandwidth.
Tissue Harmonic Imaging

59. Desired Characteristics of Contrast Agents
Efficient backscattering.
Small size for pulmonary transport.
Long half-life.
Low toxicity.
Possibility of attenuation contrast.
Possibility of speed of sound contrast.
Tissue Harmonic Imaging

60. Imaging Consideration
Low peak acoustic amplitude.
Low average acoustic power.
ECG triggering.
Frequency control.
Tissue Harmonic Imaging

61. Tissue Harmonic Imaging
System Requirements
Similarities to tissue harmonic imaging:
Minimal harmonics on transmit.
Maximal fundamental suppression on receive.
Configurable beamformer.
Wide transducer and system bandwidths.
Alternate phasing and pulse inversion may both be applicable.
Tissue Harmonic Imaging

62. Tissue Harmonic Imaging
System Requirements
Differences from tissue harmonic imaging:
Low harmonic generation during propagation, i.e. low fundamental amplitude through a small depth of field.
The fundamental amplitude may be significantly lower than regulatory limits.
Tissue Harmonic Imaging

63. Tissue Harmonic Imaging
System Requirements
Transmitter: minimal harmonic signals.
Propagation: minimal harmonic generation.
Receiver: maximal fundamental rejection.
Tissue Harmonic Imaging

64. Tissue Harmonic Imaging
System Requirements
Adequate dynamic range.
Configurable beam former.
Sufficient system bandwidth.
Wide transducer bandwidth.
Tissue Harmonic Imaging

65. Tissue Harmonic Imaging
Imaging Techniques
The phasing pattern across the array can be varied to reduce the signals at a certain frequency.
Alternate phasing is applicable to contrast harmonic imaging.
Alternate phasing on transmit is not ideal.
Tissue Harmonic Imaging

66. Tissue Harmonic Imaging
Imaging Techniques
Alternate phasing for receive: p
Alternate phasing for transmit
0.5p
Tissue Harmonic Imaging

67. Tissue Harmonic Imaging
Imaging Techniques
Difference imaging technique: contrast agents can be viewed as a “modulator”. When two different frequencies, say f1 and f2, are used on transmit, the contrast agents generate f1+f2 and f1-f2.
f1 and f2 can be properly chosen to increase the rejection ratio.
Tissue Harmonic Imaging

68. Tissue Harmonic Imaging
Imaging Techniques
Subtraction mode: Two pulses with different signs are transmitted consecutively. By adding two images together, linear components cancel while second order components remain.
This technique is susceptible to motion artifacts.
Tissue Harmonic Imaging

69. Pulse Inversion Doppler
An extension for the subtraction mode to Doppler imaging, i.e., pulse inversion with multiple firings.
There exists a nonlinear Doppler spectrum completely separate from the linear spectrum.
Potential increase in agent to tissue contrast.
Tissue Harmonic Imaging

70. Problems in Flow Estimation
Radiation force tends to move the contrast agents to a direction independent of the flow direction.
Bubble break down causes image artifacts.
Excessive backscatter produces “color blooming”.
Spectral broadening at high acoustic pressures.
Tissue Harmonic Imaging