Intelligent Analysis Of MR-HOCM Patients's Continuous Wave Doppler Spectra

In 1958,Teare first described Hypertrophic Cardiomyopathy(HCM).In clinical applications,the maximum instantaneous peak velocity of the Left Ventricle O utflow Tract(LVOT)was measured using the Transthoracic Color Doppler Echocardiography(TCDE)to calculate the LVO TG value.According to the LVOTG value in resting state,HCM will be divided into hypertrophic obstructive Cardiomyopathy(HOCM)()and hypertrophy non-obstructive type().When patients mixed with other lesions such as Mitral Regurgitation(MR),the junior physician is prone to overestimate the maximum instantaneous peak velocity and LVO TG value,affecting the optimization of disease-making and treatment decision difficult.Therefore,based on the previous research work of the Fourth Affiliated Hospital of the Fourth Military Medical University Xijing Hospital HCM research group and our lab,this paper is committed to the intelligent analysis of Continuous Wave Doppler Spectrum(CWDS)of LVOT(CWDS-LVOT),collected from HOCM patients with MR(MR-HOCM),to identify whether the LVO TG value of MR-HOCM patients are overestimated and to automatically extract the characteristic parameters of CWDS to provide richer clinical information for the hemodynamics of LVOT.In order to accomplish the above objectives,this thesis mainly made in-depth research on following aspects:(1)Collection and preprocess of CWDS-LVOT of MR-HOCM patientsThis part contains depth study of anatomical structure of the heart,acquisition and parameters extraction methods of CWDS-LVOT of MR-HOCM patients,the measurement,accuracy and possible reasons of errors of LVO TG.With the help of a junior ultrasound diagnostic physician and a senior physician with more than 20 years of experience in ultrasound diagnostic from ultrasound department in the First Affiliated Hospital Xijing Hospital of the Fourth Military Medical University,CWDS-LVOT of 12 MR-HOCM patients were collected.All LVOTG values were measured and confirmed by the two physicians.Reasonable noise of CWDS-LVO T of MR-HOCM patients were analyzed and Median filter,Mean filter and Gaussian curvature filter(GCF)were used to improve the signal noise ration.At the same time,filter effects are compared and the result indicates that GCF was the best.(2)Blind Source Separation(BSS)Algorithm and Its Applicat ionTry to make in-depth study of BSS,its implementation and the main algorithms.And understand the blind uncertainty and its reasons,especially the amplitude uncertainty and the order uncertainty.Signal separation of CWDS-LVO T of MR-HOCM patients based on FAST-ICA and reasonable blind uncertainty elimination algorithm for flow velocity and LVO TG value were designed and achieved.Estimated flow velocity and LVOTG values were compared with the results of expert diagnosis.And the sensitivity and specificity index were used to complete the accuracy evaluation of the algorithm.At the same time,an algorithm evaluation model based on Subspace discriminant ensemble classifier is established.(3)Characteristic parameters extraction of CWDS-LVOT of MR-HOCM patientsTry to study the characteristic parameters,extraction methods.The maximum frequency curve was extracted by single-degree-of-freedom model and used to extract the characteristic parameters.The extracted parameters includ es the maximum systolic velocity S,the lowest velocity of diastolic velocity D,the systolic diastolic velocity ratio SD,LVO TG,resistance index RI,pulsatility index PI,contraction spectrum width W and shrinkage rise time T.And the parameters were used to complete the assessment of blood flow conditions.(4)System interface designA simple analysis and identification system for CWDS-LVOT of MR-HOCM patients was designed by MATLAB GUI.The operation interface is divided into three parts,which are LVO TG overestimation judgment,feature parameters extraction and diagnosis analysis report.In the part of LVO TG overestimation judgment,comparison Fig of maximum frequency curve and velocity corrected blind source separation results and estimated source signal are showed.The feature parameter extraction part is to write the callback function of the corresponding control and to extract the specific parameters.The diagnostic analysis report part is to generate a report in word,which contains the extracted diagnostic parameters,patients' information and the analysis result of the overestimation judgment.In this thesis,we focus on the research of CWDS-LVOT of MR-HOCM patients and establish a simple analysis system of LVO T flow velocity by using CWDS-LVO T of MR-HOCM patients.The analysis system was used to determine whether or not the CWDS-LVOT of MR-HOCM patients were measured overestimated.In this study,the automatic extraction of the partial characteristic parameters of CWDS-LVOT of MR-HOCM patients was achieved.The effectiveness and feasibility of the system were further verified by experiments.