Characterization Presentation Performed by: Ron Amit Supervisor: Tanya Chernyakova Semester: Spring Sub-Nyquist Sampling in Ultrasound Imaging
Ultrasound Device: 2
Problem : Modern devices require large number of receivers Acoustic pulses are of high bandwidth Typical Nyquist rate is 20 MHz * Number of receivers Large amount of data must be processed High computational cost 3
4 Solution : Reduce sample rate, while still extracting the same required information for image reconstruction
FRI Model: 5
Single receiver solution : Unknown parameters are extracted from low rate samples. 6
Multichannel Sampling Scheme : Different sampling scheme for a single receiver, using bank of integrators 7
Problem : Low SNR of received signal at a single receiver. Solution : Use array of receivers and combine the received signals – Beamforming process. Beamformed signal has improved SNR Represents reflections from a single angle – forming an image line 8
Beamforming : 9
Compressed Beamforming : Combines Beamforming and sampling process. Received signals are sampled at Sub-Nyquist rate The scheme’s output is a group of Beamformed signal ‘s Fourier coefficients Digital processing extracts the Beamformed signal parameters 10
Using modulation with analog kernels and integration First Scheme : Problem : Analog kernels are complicated for hardware implementation 11
Simplified Scheme : Based on approximating each received signal by only Ki Fourier coefficients Each received signal is filtered by a simple analog filter Linear transformation on the samples provides the Beamformed signal Fourier coefficients 12
13 Analog Processing Sub – Nyquist Sampling Receiver Elements Low Rate Samples Digital Processing Amplitudes and delays of reflections Image Reconstruction Block Diagram :
14 Project Goals : Main goal: Prove the preferability of the Xampling method for Ultrasound devices Sub goals: Alternative image reconstruction Optimize algorithm and improve runtime Explore hardware implementation
15 Semester 1: Understand and run current code Improvement: Image construction from pulses Lighter OMP algorithm Semester 2: Algorithm optimization: Flow graph algorithm Complexity analysis of subroutines Runtime optimization System analysis : How to implement on processer platform for maximal performance Mission Plan:
16 תכנית עד מצגת אמצע לימוד של שיטות סטנדרטיות לייצוג תמונה מ -beamformed data ייצוג אלטרנטיבי של תמונה מתוצאות קיימות ( השיות ואמפליטודות ) השוואה בין תוצאות שמתקבלות בשתי שיטות, fine tuning של שיטה אלטרנטיבית לימוד תיאורטי של אלגוריתם ה -OMP, ניתוח של האלגוריתם שמומש אצל נועם 28.6 – 27.7 תקופת בחינות 27.7 – 1.8 הכנה למצגת אמצע