Adaptive matched filters for contrast imaging

(1)

HAL Id: hal-01075515

https://hal.archives-ouvertes.fr/hal-01075515

Submitted on 17 Oct 2014

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Distributed under a Creative Commons Attribution - ShareAlike| 4.0 International License

Adaptive matched filters for contrast imaging

Sébastien Ménigot, Iulian Voicu, Anthony Novell, Mélouka Elkateb Hachemi Amar, Jean-Marc Girault

To cite this version:

Sébastien Ménigot, Iulian Voicu, Anthony Novell, Mélouka Elkateb Hachemi Amar, Jean-Marc Gi-rault. Adaptive matched filters for contrast imaging. IEEE. Ultrasonics Symposium (IUS), 2010 IEEE International, Oct 2010, San Diego, United States. IEEE, pp.1728 - 1731, 2010, �10.1109/ULT-SYM.2010.5935573�. �hal-01075515�

(2)

Adaptive Matched Filters for Contrast Imaging

Sébastien Ménigot, Iulian Voicu, Anthony Novell, Melouka Elkateb Hachemi Amar and Jean-Marc Girault

Université François Rabelais de Tours, Tours, France INSERM, U930, Tours, France

CNRS, ERL 3106, Tours, France

October 11-14, 2010 1. Introduction

Conventionnal ultrasound contrast imaging systems use a fixed transmit frequency. However it is known that the insonified

medium (microbubbles) is time-varying and therefore an

adapted time-varying excitation is expected. We suggest an

adaptive imaging technique which selects the optimal transmit frequency that maximizes the contrast tissue ratio (CTR). This ratio can be maximized if the microbubbles backscattered

power is maximized. CTR ₌ Ebubbles E_tissue , with   

E_bubbles the microbubbles backscattered power

E_tissue the tissue backscattered power Two matched filters (MF) techniques are used to improve the image contrast. The first technique is an adaptive MF

technique and the second is a RLS technique derived from identification theory.

2. Materials

Microbubbles

◮ Microbulles SonoVueTM: mean diameter of 4.5 µm with shell

thickness of 1 nm ; Resonance frequency : f_R = 2.1 MHz

◮ Concentration: 1/2000 diluted solution of SonovueTM ◮ Immersed in a blood mimicking fluid (BMF)

Acoustical Measurements

◮ Arbitrary function generator piloted by

Matlab R ◮ 2 perpendicular transducers : 1. Emission : 2.25 MHz - BW 74% 2. Reception : 3.5 MHz - BW 63% Emission Reception 3. Methods 1. Switch in position 1

1.1 Sending a pulse on the bubbles

1.2 Receiving the backscattered signal

1.3 Repeating (1.??) and (1.??) 20 times

1.4 Linear combination(ACP) from this 20 signals 1.5 Identification system with an adaptive filter

2. Switch in position 2

2.1 Amplification of the signal such as E₅ ₌ E₂ and reversal 2.2 Pulse is the signal

2.3 return to ??

4. Matched Filter

First Matched Filter

2 1 x(t) y(t) E_y = E_x ˆ y(−t) Transducer 1 Bubbles Transducer 2

Recursive Least Squares (RLS) filter

Identification by RLS, i.e. minimizing the squared error such as

e_n = y_n − x_nT θ_n₋₁ Bubbles Transducer 1 Transducer 2 -Model RLS (θ) x(t) y(t) e(t) ˆ y(t) ˆ y∗(−t) E_y_ˆ = E_x 1 2

5.1 Results & Discussion : first simulation

0 2 4 6 8 10 12 14 16 18 20 −150 −100 −50 0 50 100 150 Time (s)

Pressure level (kPa)

First Pulse (1) 60 65 70 75 80 60 65 70 75 80 60 −1 −0.5 0 0.5 1 Time (µs)

Pressure Level (Pa)

Backscattered signal and its fourth−order model (2)

60 62 64 66 68 70 72 74 76 78 80 −150 −100 −50 0 50 100 150 Time (µs)

Pressure Level (kPa)

Optimal pulse (3) 60 62 64 66 68 70 72 74 76 78 80 −1 −0.5 0 0.5 1 1.5 Time (µs)

Optimal pulse (4)

Gain between signal in (fig 1) and signal in (fig 4) : 17.68 dB

5.2 Results & Discussion : second simulation

0 2 4 6 8 10 12 14 16 18 20 −100 −50 0 50 100 Time (µs)

Optimal pulse with white noise

0 2 4 6 8 10 12 14 16 18 20 −0.4 −0.2 0 0.2 0.4 Time (µs)

Pressure Level (Pa)

Backscattered Signal with Noise Optimal Pulse

SNR = 4.74 dB

Gain with RLS : 24.72 dB

Gain without RLS : 10.85 dB

A Filter RLS of fourth order is sufficient to modelize the bubble system. The RLS filter allows us to extract the information of the bubble system better than the simple match filter.

5.3 Results & Discussion : experiment

2 4 6 8 10 12 14 16 18 20 0 1 2 3 4 5 6 7

Experiment with Matched Filters

Iterations

Gain (dB)

Matched filter RLS technique

Compared to the non-optimized imaging technique, the both proposed methods give a gain superior to around 5 dB.

◮ Advantage: gain improvement ;

◮ Drawback: increasing of the system complexity.

6. Conclusion & future prospects

◮ Optimization works well even though for nonlinear system ; ◮ Implementation in an open echograph.

E-Mail: sebastien.menigot@etu.univ-tours.fr jean-marc.girault@univ-tours.fr

Acknowledgement: This work was supported by Agence Nationale de la Recherche.