The Application Of Optical Flow Algorithm Based On Affine Model In Estimation Of Organizational Motion

Objective: Ultrasound elastography is a new imaging modality developed in recent years.It can quantitatively reflect the bioelastic information of tissues and make up for the shortcomings of traditional imaging methods,which has important clinical value for early diagnosis of tumors and judgment of tissue fibrosis.The traditional ultrasonic elastic imaging algorithm first estimates the displacement,and then obtains the strain field according to the displacement gradient information,which results in the strain imaging quality depending heavily on the displacement accuracy.Most of the existing ultrasonic elastography algorithms only consider translation motion,neglecting the deformation of tissue,resulting in inaccurate estimation of displacement field especially in the case of large displacement.Therefore,the main purpose of this paper is to propose a more efficient ultrasound elastic imaging algorithm for the above problems.Methods: Affine transformation is a comprehensive form of motion,which has both displacement and deformation.Optical flow method is a two-dimensional motion estimation algorithm with high efficiency and can get sub-pixel displacement.Therefore,this paper combines affine transformation with optical flow method to get displacement and strain information simultaneously.However,because of the constraints of its basic theory,the traditional optical flow algorithm will deviate from the actual motion when calculating complex conditions(such as noise,large displacement,etc.)directly and even make calculation errors,so the accuracy of motion estimation is greatly reduced.In view of the above problems,the idea of prior estimation is used in this paper,which mainly takes the displacement of the previous row as the initial displacement of the next row,so that the displacement can be corrected in a small range.Therefore,this idea solves the problem that the optical flow method cannot directly calculate the displacement under large compression.Results: Four kinds of compressed RF echo signals of 1%,2%,5% and 8% are calculated by Field II,which basically verify the correctness of the algorithm.Quantitative analysis of the axial strain shows that the signal-to-noise ratio of the target area are improved by 3.73 d B and 6.17 d B,and the contrast-to-noise ratio are improved by 8.53 d B and 11.90 d B compared with the 2DAM algorithm and pyramid optical flow algorithm in the case of large compression.The experimental results of bionic data further verify the correctness of the proposed algorithm.The strain changes of tissue after radiofrequency ablation can be clearly seen in the strain map estimated by the proposed algorithm.Conclusion: The simulation results show that the imaging quality of the algorithm is better under the condition of large compression.The experimental results of bionic data and clinical data show that the proposed algorithm is correct and preliminarily applicable to clinical data.