Lecture: The ultrasound imaging systemfolk.ntnu.no/alfonsom/TTK13/lib/Lecture_FrontEnd_2014.pdf · The compact ultrasound imaging system • Portable lap-top sized scanner •Weight

Lecture: The ultrasound imaging system

The high-end ultrasound system from the outside

• High-quality screen

• Stereo speakers :-)

• Lots of buttons

• Small printer

• Probe connectors

• Input for ECG and auxillary pressure, and more

• Optical storage (VCR is finally out)

The high-end ultrasound imaging system

• Portable within buildings

• Weight ~ 200kg

• Multiple probe support

• Phased-array, (curved) linear-array, 2-D arrays, transesophagael (TEE)

• A range of imaging modalities

• 1-D / 2-D / 3-D / 4-D B-mode and Doppler imaging, duplex, triplex

• State-of-the art image quality and features

• Coded excitation, compound scanning, post-processing, real-time 3-D imaging

• High connectivity

• Patient reporting and data storage• PACS / DiCOM support• Support for legacy storage formats

Sequoia 512 iU2

2

Logic 9

Vivid E

The compact ultrasound imaging system

• Portable lap-top sized scanner

• Weight ~ 5-10 kg• Battery operated

• Close to high-end functionality

• Multiple probe support

• Phased-array, (curved) linear-array, TEE

• Several modalities

• 2-D B-mode and Doppler imaging, duplex and triplex capabilities, 3-D (mechanical)

• Connectivity

• Patient reporting and data storage

• PACS / DiCOM support

Cypress

The hand-carried ultrasound system

• Very portable• Weight ~ 1-3 kg

• Simple functionality

• Fast boot-up time

• Varying probe support• Phased-array, (curved) linear-array

• The most important modalities• 2-D B-mode and color-Doppler

imaging

• Limited connectivity• Frame buffer

• Data transfer

Sonosite iLook

(2002)

SonoHeart Elite

(2002)

Philips OptiGo

(2001)

Ultrasound minaturization

• The miniaturization of ultrasound systems has mostly followed the tremendous development of integrated circuits and power efficient electronics

1985 2000 2009 - 2015

Logic 9

SSD-880CW

FremtidenVivid E

2008

Medical ultrasound imagingModalities

Time

De

pth

M-mode (motion) B-mode (brightness)

Medical ultrasound imagingModalities

PW-Doppler Color-DopplerH

as

tig

het

Tid

Building blocks of a generic ultrasound imaging system

• Transducer

• Converts electric signals to acoustic pressure waves and vice versa• Front-end

• Controls the transmit and receive beamforming and further signal formation• Mid-processors

• Hardware (DSP) signal processing and image formation, Doppler processing (CW)• Trend towards integration with the back-end system

• Back-end

• Post-processing, image formation, data storage and user interfacing

The ultrasound transducer

Current high-end technology

• Arrays of piezoelectric elements

• Focusing and steering by delaying transmission on individual elements

• Composite cheramics (PZT, Lead Zirconate Titanate)• Needs absorbing backing and impedance matching layer to

achieve a sufficient bandwidth for high-resolution imaging• Limited bandwidth (< 80 % relative to center frequency)

PZT Substrate Backing Layer

Matching Layer Electrode

. . .

Transducer surface

Annular array

~ 8 elements

1-D linear array

64-192 elements

Multiplexing

Full 2-D linear array

~ 2000 elements

1.25D-1.75D linear array

64-128 * rows elements

Multiplexing / symmetry

Transducer topologies:

Transducer shape and function

• Ultrasound transducers are optimized for a given frequency range

• Due to the fundamental resonance modes

• Due to frequency dependent attenuation

Transducers come in different shapes

• The acoustic window varies with application

• Typical applications

• Cardiac: phased-array, 1-5 MHz

• Abdominal: curvilinear array, 2-7 MHz

• Central vessels: Linear array, 5-10 MHz

• Peripheral: Linear array, 8-15 MHz

• Intravascular (catheter) > 20 MHz

Capacitive micromachined ultrasound transducer (cMUT)

• A capasitive membrane oscillates in response to an external electric field / mechanical force

• Pros

• Can be integrated onto silicon wafer• Have a high bandwidth (>100%) for

multi-purpose operation• Elements can be made very small

(~10-100 μm)• Cons

• Not yet as sensitive as PZT transducers

• Generate higher harmonics on transmission, preventing second harmonic imaging

• Cross-talk issues

cMUT schematic (right) and experimental array with flexible interconnect circuit (left).

Imaging of the carotid artery using GE 12L linear array (right), and an experimental cMUT array (left)

Source: Mills D.M. Medical imaging with capasitive micromachined ultrasound transducer (cMUT) arrays. IEEE Ultrasonics symposium, Rotterdam, 2004

N

Main front-end components (digital)

• Transmit pulse generator• High-voltage transmitters (50 - 100V)

• High-end system: Arbitrary waveform generators (D/A conversion)

• Pre-amplifiers and analog-to-digital converters• High-end system: Low-noise depth-dependent amplifier, high-speed (~40MHz) 12 bit A/D converters

• Receive beamforming• High-end system: Fully digital time-delay beamformer, dynamic focus and apodization

• Demodulator• High-end system: Discrete Hilbert transform

Transd

ucer

T/R switches

Tx Amp Transmit pulse generator

ADC Receive beamformer

Depth-dependent Pre-amplification

Analog-to-digital conversion

DemodulatorNN 1

Coherent summation of channels

Transformation to baseband

N

N ASIC / DSP / FPGA / GPU

Transmit/receive switch, protection

TGC

Mid-processors

• Special purpose hardware components for signal processing and image formation

• Application specific integrated circuits (ASICs)

• Digital signal processors (DSPs)

• Field programmable gate arrays (FPGAs)

• Typical tasks:• Computationally intensive signal processing

• Image formation processing

• Doppler processingFigure: Mid-processors, special purpose hardware for signal processing and image formation

The rapid improvement in computational power of general purpose CPUs has moved an increasing number of mid-prosessing tasks to software in the back-end system

Back-end system

• Implemented using general purpose CPUs and systems

• Controls an increasing number of tasks:

• Software-based Doppler processing

• Post-processing of ultrasound data

• Image scan conversion

• Data storage and connectivity

• Latest developments• GPU (parallel) processing

• Software beamformingFigure: Plans for the Intel IA-32 system-on-chip. Dimensions ~ 4/4/0.4 cm (h/w/d), Pentium M processor, 600-1200 MHz, 13-22 W power consumption

Transmit pulse generators

• State-of-the-art: Arbitrary waveform generators

• High-voltage transmitters (50 - 100V)• Digital-to-analog convertion of arbitrary waveforms• Allows for advanced coded pulse excitation• Allows for advanced transmit apodization

• Design challenges

• High-voltage transmitter needed• One D/A converter / high-voltage transmitter for each channel• Consumes power and space

• Typical trade-offs:

• Resolve to (bipolar) square-wave transmission• Reduce voltage to reduce to power consumption => reduced

sensitivity

Image quality requirement:

Use sufficient amplitude for second-harmonic imaging

T/R switches

Tx Amp Transmit pulse generator

N

N

Transd

ucer

Receive chain

Receive electronics – one channel

• Low noise amplifier brings received signal (mV) to standard levels

• A variable gain amplifier is needed to equalize the signal from different depths(due to depth/frequency dependent attenuation), i.e. reduces dynamic range

• Anti-aliasing filter with cut-off ~20MHz (often integrated in the ADC)

• Analog-to-digital conversion at >40MHz, and typically 4x upsampling

LPF ADCLNA VGA

Transducer element

Anti-aliasing filter

A/D converter

Low-noise amplifier

Variable gain amplifier

Beamformer

Pre-amplification Analog-to-digital conversion

.

.

.

The dynamic range in ultrasound imaging

• The dynamic range of the received ultrasound echoes are influenced by

• Reflection losses, 20-30 dB, frequency dependent attenuation 1 dB/MHz·cm (pulse-echo)

• Example: Cardiac imaging• Freq. 2 MHz, depth 20 cm => 60-70 dB dynamic

range• Second-harmonic imaging => ~ 80-100 dB

Implications:

• Low-noise and depth dependent pre-amplifiers are needed to reduce the dynamic range

• High-resolution A/D-converters needed (at least 12 bit)

Depth

Transducer

Scattering sources

Source: Schafer M. E. and Lewin P. A. The influence of front-end hardware on digital ultrasonic imaging. IEEE Trans. Son. Ultrason., vol. 31, pp. 295-306, 1984

Channel count vs. image quality

• The channel count is proportional to the number of aperture elements

• Smaller elements are desired to increase the fundamental lateral resolution, and to avoid grating lobes

• Larger apertures are desired to obtain a narrow focus and to increase sensitivity

Annular array

~ 8 elements

1-D linear array

64-192 elements

1.25D-1.75D linear array

64-128 * rows elements

Transducer topologies:

Full 2-D linear array

~ 2000 elements

A/D

.

.

.

No

. of

chan

nel

s

Aperture vs. beam width

Depth

=> Trade-off between channel count and image quality

Channel count vs. complexity

• A high-voltage transmitter needed for each transmit channel

• A low-noise depth-dependent pre-amplifier needed for each receive channel

• An analog-to-digital converter needed for every digital channel

• The complexity of the beamformer increases with the number of channels

LPF ADCLNA VGA

Transducer element

Anti-aliasing filter

A/D converter

Low-noise amplifier

Variable gain amplifier

Beamformer

Pre-amplification Analog-to-digital conversion

.

.

.

Minimize / trade-off the channel count due to: Cost, power consumption, miniaturization

Reducing the channel count / complexity

• Increasing the element size• Reduce lateral resolution, increase element directivity, grating lobes issues

• Multiplexing• Select a subset of the available elements used, applied for instance in linear arrays

• Sparsed array imaging• ”Optimally” select a subset of elements for beamformation

• Analog beamforming• Reduce the complexity of the beamformer

• Hybrid beamforming• Analog beamforming of smaller subapertures, before A/D and digital beamforming

• Synthetic aperture imaging• Transmit / receive on small subapertures successively across aperture, synthesize beamforming

• Delta-sigma A/D conversion and beamforming• One-bit high-rate sampling and beamforming

Receive beamforming

• Steering and focusing of the beam through the summation of the signal from each array element

• High-end systems: A fully digital delay-and-sum beamformer

Digital or analog receive beamforming

• Advantages of analog beamforming

• Only one high-resolution ADC is needed

• Analog beamformers reduce cost and take less space

• Lower power consumption

• Disadvantages of analog beamforming

• Number of delay taps is limited

• Analog delay lines tend to be poorly matched channel-to-channel

• Advantages of digital beamforming

• More accurate signal summation with less harmonic distortion

• Parallel beamforming

• Integration

Digital beamforming:

Analog beamforming:

Digital beamformer needed for high-end image quality

(Second-harmonic imaging, parallel beamforming…)

Hybrid beamforming

• Combine analog and digital beamforming

• Example: Fine analog pre-beamforming of neighboring elements prior digital beamformer

• Digital beamforming benefits on combined channels

• Patented for use in 2-D array systems to reduce channel count and cables

ABF

ABF

ABF

DFE

Analog beamformers

Digital beamformer

Back-end

Scattering source

Source: Petrofsky et al. Ultrasonic receive beamformer with phased sub-arrays. US Patent 5,573,001, 1996

Integrating the ultrasound transducer and front-end electronics

• Motivation – ”probe-on-chip”

• Minimize cable and connector size, wireless?• Increase signal-to-noise ratio• Streamline fabrication process

• Current piezo-electric transducers

• Needs absorbing backing• Not compatible with integrated circuit

manufacturing• Needs flex-connection, and wiring

On-chip integration of piezo-electric transducers not possible

However, trend towards moving front-end to the probe handle

. .

Tran

sdu

cer

surf

ace

System front-end

T/R VGA

Tx TPG

ADC Beamf.

Demod

N N1

N

NCable

Pro

be

han

dle

Pro

be

con

nec

tor

Flex

co

nn

ecti

on

System front-end

T/R VGA

Tx TPG

ADC Beamf.

Demod

N N1

N

N

. .

Tran

sdu

cer

surf

ace

Cable

Probe handle

Synthetic aperture imaging

• Transmit and / or receive on single elements or subapertures successively across an aperture

• Reconstruct a high-resolution image by combining several low-resolution images

• Pros:• Reduced number of high-voltage transmit channels and / or receive channels

• An increased frame rate and dynamic focus on both transmit and receive can be achieved

• Cons:• Reduced sensitivity and spatial resolution

• Suceptible to motion artifacts

• Computationally demanding with large memory requirements

Sources:

Karaman et al. Synthetic Aperture Imaging for Small Scale Systems. IEEE Trans. on Ultrason., Ferroelect., and Freq. Contr., vol. 42, May 1995

Kim J-J. and Song T.K. Real-Time High-Resolution 3D Imaging Method Using 2D Phased Arrays Based on Sparse Synthetic Focusing Technique. IEEE Ultrasonics symposium proceedings, 2006

Coherently combine images from several low resolution emissions to produce final image