Pixel-based Beamforming for Ultrasound Imaging

Transcription

1 Pixel-based Beamforming for Ultrasound Imaging Richard W. Prager and Nghia Q. Nguyen Department of Engineering

2 Outline v Introduction of Ultrasound Imaging v Image Formation and Beamforming v New Time-delay Calculation v Results v Conclusions

3 Outline v Introduction of Ultrasound Imaging v Image Formation and Beamforming v New Time-delay Calculation v Results v Conclusions

4 Ultrasound Imaging Imaging (RF) data US scanning B-mode FPL Acquisition APB Demodulation Scanning flexor tendon using ultrasound FPL = Flexor Pollicis Longus tendon APB = Abductor Pollicis Brevis muscle v Pulse-echo Imaging, using acoustic wave/pressure to form images v Frequency range: 1-4 MHz v Many applications imaging soft tissue in clinical medicine (not good for air or bone)

5 Outline v Introduction of Ultrasound Imaging v Image Formation and Beamforming v New Time-delay Calculation v Results v Conclusions

6 Image Formation: Dynamic Focusing RF data Dynamic focusing Delay and Sum Wavefront alignment τ 1 echoes τ 2 Single transmit beam Summed RF data Σ τ 3 τ 4 scatterer τ 5 Apodization Delays v Image is generated line-by-line using fixed focused transmit and dynamic receive focusing image resolution is only optimal around focal depth, (transmit focal depth problem). v Synthetic scan-lines are interpolated between actual scan-lines, (interpolation problem).

7 Beamforming Transducer Store the raw received data from all the transmitted beams. This enables us to solve both problems: Fix interpolation problem: perform individual focus calculations for each pixel. Fix the single transmit focal depth problem: combine information from several beams to improve the focussing away from the focal depth. To focus using data away from the centreline we need the travel times for these points.

8 Time-Delay Calculation: Conventional Approach Linear Array x (x i, z i ) (x r, z r ) z = (x p x i ) 2 +(z p z i ) 2 c + (x p x r ) 2 +(z p z r ) 2 c (x p, z p ) c Sound speed Ø Ø Ø Using geometrical optics approximations Assuming that the transmitted pressure field is spherical and generated from the center of the active aperture Used in convex array systems with data compounding limited to two transmit beams (Lee et al., IEEE Trans. UFFC, 59(3): , 212.)

9 Time-Delay Calculation: Virtual Source Approach Linear Array v A virtual source is assumed at the focal point. This generates a spherical wave. v The assumption is valid only within a limited angle (shown by the dotted lines). Imaging data outside the angle is currently discarded. v Creates artefacts around the focal depths because of the discontinuity in the spherical wave approximation. (x i, z i ) d a (x p, z p ) (x r, z r ) v Transmit beams can be made more broad with the focal depth outside the imaging region (SA-BiPBF: Kim et al., IEEE TBE, 6(1): , 213.) i = d ± a c + (x p x r ) 2 +(z p z r ) 2 c d focal length

10 Outline v Introduction of Ultrasound Imaging v Image Formation and Beamforming v New Time-delay Calculation v Results v Conclusions

11 Simulated Pressure Waveforms v Goal: Maximize data compounding at individual pixels. 5.6 Pressure fields (norm.) A B C µs at A at B at C v v The virtual source approach is only valid in some places. A better approximation is to describe the beam at each point in terms of two pulses of varying amplitude.

12 Field Pattern Analysis Linear Array 5 1 max min conven. Axial (mm) Time (µs) Nguyen and Prager, High-resolution ultrasound imaging with unified pixel-based beamforming, submitted to IEEE TMI (May 215).

13 New Time-Delay Calculation (I) system delay 1.5 Pressure field (norm.) at A R min d a A µs 1 15 τ delay between the transmit time at the edge and centre of the active aperture. trans = R min c = d a c

14 New Time-Delay Calculation (II) system delay 1.5 Pressure field (norm.) at C d R max a µs 2 25 C trans = R max c = d + a c

15 New Time-Delay Calculation (III) system delay 1.5 Pressure field (norm.) at B d R max B 1 R min B µs 15 2 B 2 min = R min c and max = R max c trans = dist(b 2,B) dist(b 2,B 1 ) min + dist(b 1,B) dist(b 2,B 1 ) max

16 Outline v Introduction of Ultrasound Imaging v Image Formation and Beamforming v New Time-delay Calculation v Results v Conclusions

17 Demonstration Beamformers Focal depth No. of beams per scanline Dynamic Focusing 2 mm 1 Conventional Pixel-based 2 mm 64 and 8 SA-BiPBF 4 mm 64 Unified Pixel-based (proposed method) 2 mm 64 All images have a depth range from 3 mm to 34 mm

18 Simulated Results 5 Point targets Dynamic focusing Con. PB (64) SA-BiPBF(64) Unified PB (64) Axial (mm) Lateral (mm) -4 4 Lateral (mm) -4 4 Lateral (mm) -4 4 Lateral (mm) -4 4 Lateral (mm)

19 Lateral Profiles 1 At 5 mm 1 At 12.5 mm Power (norm.) [db] -1-3 Power (norm.) [db] -1-3 Power (norm.) [db] At 2 mm 1 Dyn. Con. PB SA-BiPBF -1 Uni. PB -3 Power (norm.) [db] At 27.5 mm Lateral (mm) Lateral (mm)

20 Phantom scanning Transducer

21 Phantom results Dynamic focusing Con. PB (64) Con. PB (8) SA-BiPBF(64) Unified PB (64) 5 Axial (mm) Lateral (mm) Lateral (mm) Lateral (mm) Lateral (mm) Lateral (mm)

22 In vivo results Dynamic focusing FPL Conventional PB (8) Unified PB APB FPL = Flexor Pollicis Longus tendon APB = Abductor Pollicis Brevis muscle

23 Outline v Introduction of Ultrasound Imaging v Image Formation and Beamforming v New Time-delay Calculation v Results v Conclusions

24 Conclusions v Good performance of the proposed unified PB beamforming comes from the highly focused beam on transmit and dynamic time delay calculations on receive. v The time-delay calculation still uses the geometric optics approximation which makes for simple implementation but limits us to the use of only one data point per received waveform. v The method leads to enhancements in image quality in both phantom and in vivo studies. Acknowledgements: Dr Laurence Berman for performing and interpreting the in vivo scans.