The Value Of Magnetic Resonance Deformation Analysis In Evaluation Of Myocardial Function

Part 1 Myocardial deformation analysis:comparison of cardiac magnetic resonance feature tracking and deformable registration algorithmsObjectives:To evaluate deformable registration algorithms(DRA)-based quantification of cine steady-state free-precession(SSFP)for myocardial strain analysis in comparison with feature-tracking(FT)Materials and Methods:We recruited 40 healthy volunteers(60%male,mean age 49.50±8.48 years)between September 2018 and November 2018.All volunteers underwent cardiac magnetic resonance imaging on a Philips Ingenia 3.0T magnetic resonance scanner.The scanning sequence mainly includes short-axis,two-chamber,three-chamber and four-chamber SSFP cine imaging and short-axis SSFP cine sequence after contrast injection.Using the DRA(TrufiStrain version 2.0,Siemens Healthcare,A.G.,Germany)and FT(Mass v2.0,Leiden,Netherlands)algorithm-based myocardial strain analysis software to evaluate the left ventricular global radial strain peak(GPRS),endocardial global toroidal strain peak(endoGPCS),myocardial middle layer overall circumferential strain peak(midGPCS)and epicardial global circumferential strain peak(epiGPCS)and basal,left ventricular intermediate and apical circumferential strain peak(PCS)and radial strain peak(PRS).The reproducibility of myocardial strain analysis based on plain and enhanced cine images was compared by calculating the intra-class correlation coefficient(ICC)and coefficient of variation(COV)and performing Bland-Altman analysis and Pearson correlation analysis.Results:① Comparison of the absolute values of the two algorithms:GPRS,endoGPCS,midGPCS,epiGPCS,and PRS,PCS of basal,medial,and apical left ventricle were significantly different in myocardial deformation analysis based on plain or enhanced short-axis cine images.The corresponding strain parameters of FT were significantly greater than that of DRA(P<0.05).②The effect of contrast agent on the measured values of the two algorithms:the correlation of strain parameters measured based on between plain and enhanced cine images was significantly better when using DRA algorithmsthan FT.③The differences of peak strain values based on DRA between plain and enhanced cine images were larger than those based on FT algorithm ④ The reproducibility of the two algorithms:regardless of the plain or enhanced cine images,the ICC of the strain parameters of the DRA measurement is larger and the COV is smaller.According to results of the Bland-Altman analysis,the average circumferential parameters of each layer and each part measured by the DRA algorithm and the average bias and/or uniformity limit range of the radial strain of the whole and each part are smaller than the corresponding results measured by the FT algorithm.⑤Effect of contrast agent on measurement reproducibility of two algorithms:Compared with enhanced film images,strain parameters measured through DRA algorithm has larger ICC and smaller COV among observers when based on plain cine images,while FT algorithm performs better when based on enhanced movie images.Conclusions:Compared to the FT algorithm,the DRA analysis algorithm based on pixel point analysis has lower variability and excellent inter-observer and intra-observer measurement repeatability.In addition,contrast injection has a significant impact on the strain peaks measured by the two algorithms,especially the DRA algorithm.In clinical applications,attention should be paid to maintaining the consistency of the acquisition method and time of cine images.Part 2 Left Ventricular Myocardial Deformation Analysis:A study on diastolic function in Chinese male population and its relationship with fat distributionsObjective:To explore prognostic value of a deformable registration algorithm(DRA)-based myocardial strain rate analysis with steady-state free-precession(SSFP)cine cardiac magnetic resonance(MR,CMR)images in measuring diastolic function in obesity and its relationship with fat distributions.Materials and Methods:A total of 136 male participants with a spectrum of body mass index(BMI)underwent CMR scan using a 1.5T MAGNETOM Aera(Siemens Healthcare,A.G.,Germany),1H-MRS hepatic fat fraction,and computed tomography(CT)scan(Somatom Definition Flash;Siemens,Erlangen,Germany).All the participants were enrolled from September 2017 to April 2018.A Trufi-Strain prototype software(version 2.0,Siemens Healthcare,A.G.,Germany)was used to quantify strain rate(SR)in both early diastole from cine images(RSRE,CSRE and LSRE for radial,circumferential and longitudinal ones respectively)and late diastole from cine images(RSRL,CSRL and LSRL for radial,circumferential and longitudinal ones respectively).Ratios of early and late SRs were calculated(REL,CEL and LEL for radial,circumferential and longitudinal ones respectively).Areas of epicardial and pericardial adipose tissue(EAT and PAT)were measured on a single 4-chamber-view slice of cine images.Volumes of visceral and subcutaneous adipose tissue(VAT and SAT)were acquired semi-automatically from CT images using the dedicated software(Cardiac Risk Assessment 2.0,Siemens).Waist and hip circumferences were manually measured(WC and HC).Analysis of variance or nonparametric tests,correlation and stepwise multivariate regression analysis models were applied for statistical analysis.Results:Only RSRE exhibited significant differences between obese and lean participants(-130.22±28.52 vs.-113.19±31.24,P=0.029)while no significant differences between groups of CSRE and LSRE was found(all P>0.05).All peak late diastolic SRs of obese men were higher than lean ones(RSRL:-37.10±10.09 vs-29.34±7.91,CSRL:66.19±18.06 vs 45.59(41.76,55.34),LSRL:55.49±13.97 vs 41.25±6.63,all P<0.001)while All SR ratios of obese subgroups were all lower than lean ones(REL:3.26±1.19 vs 4.65±1.21,CEL:2.16(1.85,2.70)vs 3.01±0.62,and LEL:1.68±0.50 vs 2.23±0.62,all P<0.001).EAT was a significant predictor of SR ratios withβ of-0.487,-0.306 and-0.165 for radial,circumferential and longitudinal directions,respectively.Conclusion:SR ratios of early and late diastole derived from cine images using DRA-based software might serve as imaging marker of early subclinical diastolic dysfunction in obesity.And EAT might be a significant predictor of diastolic dysfunction in obesity.Part3 Diagnostic and Prognostic implications of deformable registration algorithm-based myocardial deformation ananlysis in hypertrophic cardiomyopathy:Relationship with left ventricular hypertrophy and fibrosisObjective:To explore the value of myocardial strain analysis based on deformable registration algorithm(DRA)in the diagnosis of hypertrophic cardiomyopathy(HCM)and its relationship with prognostic risk factorsMaterials and Methods:We retrospectively collected 23 patients with HCM who were diagnosed in our hospital from October 2017 to October 2018 and who had undergone standardized cardiovascular magnetic resonance(CMR)imaging.There were 18 male patients and 5 female patients,aged from 30 to 79 years,with an average of(50.70±14.80)years old.In addition,23 volunteers who matched the age and sex were also recruited All volunteers underwent a CMR scan consistent with patients with HCM.The scan sequences included the left ventricular short axis,the two-chamber,four-chamber heart steady-state free-precession(SSFP)cine images,and the late gadolinium enhancement(LGE)scan.The HCM group was divided into no-enhanced subgroup and enhanced subgroup according to the presence or absence of LGE.Images of patients with HCM and volunteers were analyzed using the DRA-based myocardial deformation post-processing software named TrufiStrain.The measurement parameters included global peak radial strain(GPRS),global peak circumferential strain(GPCS),global peak longitudinal strain(GPLS),and peak radial,circumferential and longitudinal strain(PRS,PCS,and PLS)of different parts of the left ventricle(the apex,the middle of the left ventricle,and the base).Student-t test or Mann-Whitney,Person correlation analysis,ROC curve were used to statistically analyze the diagnostic efficacy of myocardial deformation parameters in HCM and its relationship with left ventricular wall thickness and LGE.Results:Other than apical PCS,all GPRS,GPCS,GPLS,basal and medial PRS,PCS,PLS,and apical PRS,PLS in patients with HCM were lower than those in volunteers(P<0.05).The global,apical,basal and medial radial,circumferential and longitudinal strain and strain rate of the patients with HCM were significantly smaller than those of the control group(P<0.05).There was no significant difference in PRS,PCS and PLS between the control group and the group without LGE(P>0.05),and significantly higher than those in the patients with LGE(P<0.05).The strain rate ratios of patients with LGE were smaller than those in the patients without LGE and the control group(P<0.05).There was a statistically significant difference of the radial strain rate ratio of medial and apical left ventricle and the circumferential rate ratio of the basal left ventricle between patients without LGE and the control group(P<0.05),while the others between the two groups were not(P>0.05).GPRS,GPCS,GPLS,REL,CEL and LEL were all significantly correlated with the maximum thickness of left ventricular myocardium(P<0.05).ROC curve analysis results showed that areas under the curve of GPRS,GPCS,GPLS,REL,CEL and LEL for the diagnosis of HCM were 0.686,0.905,0.921,0.972,0.974 and 0.917,respectively.The area under the curve GPRS,GPCS,GPLS,REL,CEL and LEL predict LGE in patients with HCM is greater than 0.871Conclusions:DRA-based myocardial deformation analysis is of great value in diagnosing HCM and predicting LGE in HCM patients and is closely related to myocardial thickness.