Research Of 2-D Ultrasound Elastography Algorithms Based On RF Signals

Cancer has been always threatening human's health,with more than 1500 thouands new sufferers every year.Research shows that many cancers don't have pathological and physiological changes in the disease early stages,but only with elastagraphy changes,such as liver fibrosis,breast cancer and so on.Common imaging techniques,such as CT,MR]and ultrasound imaging,are difficult to detect these diseses.Ultrasound elastography can detect tissue elasticity information,which can not be detected by common imaging technologies,and is widely used in clinical diagnosis.Guasi-elastography imaging technology has been widely used,because of its high resolution,real-time and with no additional ultrasound equipment.Current studies of quasi-elastography mainly focused on axial direction.However,tissue also deforem in the lateral direction.At present,researches of 2-D(two-dimentional)elastography mainly based on B-mode ultrasound image data,which contains less information compared to RF(Radio Frequency)data,which is the raw data obtained in ultrasound imaging.It is more accurate to make strain analysis based on RF data.Displacement is always estimated before strain anlysis.NCC(Normalized Cross-correlation)is generally considered to be the gold standard of motion estimation.However,NCC has a high computational cost,especially when RF signals are used.Besides NCC algorithm can't estimate large compression strain.According to the problems in quasi-static imaging techniques and the existing research results,this paper studys tissue 2-D strain estimation based on RF data.Firstly,2-D displacement was estimated based on NCC algorithm.Secondly,this paper proposed a new algorithm,predicted NCC algorithm,based on NCC method.In predicted NCC:(1)fast NCC based on sum-tables was used;(2)displacement prediction was made based on the previously calculated displacement of its neighbor window before calculating the displacement of the present window.Thirdly,NCC and predicted NCC method can only estimate small compression,thus this paper displacement estimation based on DP-AM(Dynamic Programming and Analytc Minimization)was used to estimate large compression.At last,three kinds of strain estimation method were performed in this paper,which are direct gradient,LSQSE(Least-Squares Strain Estimator)and LSQSE with Kalman Filter method,to estimate strain to obtain elastogram.Computer simulation data,ultrasound phantom data and in-vivo human dada were used in this paper to verify the algorithms respectively.And the experimental results show that:1)predited NCC proposed in this paper outperforms NCC method in computation time and noise immunity;2)displacement estimation method based on DP-AM can estimate large dynamic range compression strain;3)LSQSE with Kalman filter straion method outperforms direct gradient and LSQSE method.