Role Of Spatio-Temporal Image Correlation In Measuring Fetal Left Ventricular Function

Object. The purpose of this study was to compare the agreement and reliability of three techniques(Virtual Organ Computer-aided AnaLysis, VOCAL; Inversion mode, Inversion; and Sonography-based Automated Volume Count, SonoAVC), for measurements of fetal ventricular volumes. And to develop a semi-automatic and repeatable application of a tool for fetal ventricular volume measurements.Methods. (1)Experiments to validate the accuracy of the ultrasound system and measurement technique were performed with 10 modified jelly, within which have different size fruit blocks.10 different size fruit block model were used water method to measure the volume in the graduated cylinder. A known volume of the fruit block into the jelly to fixed the block, and form fruit block-jelly contrast interface to simulate heart chamber-myocardial interface. Volume measurements were performed independently by two observers. Intraobserver reliability was assessed by calculating the intraclass correlation coefficient (ICC), and coefficient of variation (CV), which was defined as the SD of within-subject differences expressed as a percentage of the mean as described by Bland and Altman. ICC values of >0.95 and CV<5% were regarded to reflect good reliability and agreement respectively. CV<10% was regarded to reflect acceptable agreement. (2) We studied 106 volumes from fetal hearts in singleton pregnancies at 20-32 gestation breast tumors were 85.7% and 94.8%,and the positive and negative predictive values were 94.7% and 85.9% respectively. weeks. STIC data sets were frozen in end-systole and end-diastole. 1)Ventricular volumes from 96 fetuses with normal hearts were measured with VOCAL, Inversion and SonoAVC. The stroke volume was calculated from these measurements. Reliability and agreement of the three techniques were evaluated with intraclass correlation coefficients (ICCs), and proportionate Bland-Altman plots were constructed. The time necessary to complete the measurements with either technique was compared using Student's t-test and P-values<0.025 were considered statistically significant. Intraobserver and interobserver agreement of measurements was calculated.2) The stroke volume was calculated by one observer using three techniques of 10 fetuses with cardiomegaly,and then reliability and agreement of the three techniques were evaluated, and the left ventricular ejection fraction were compared using Student's t-test and P-values<0.05 were considered statistically significant.Results. (1) Ten individual volumes in experiments ranging from 0.8 to 3.0 mL were analyzed. Volume measurements with three techniques of STIC were highly correlated (All of the mean percentages of difference were<8%, all ICCs> 0.95, and all CVs<5%). A high degree of reliability was observed between:VOCAL and inversion mode (Observer A's ICC,0.998; 95% CI,0.926-1.000; Observer B's ICC,0.999; 95% CI,0.936-1.000); VOCAL and SonoAVC (Observer A's ICC,0.990; 95% CI,0.202-0.998; Observer B's ICC,0.980; 95% CI,0.0191-0.997); and inversion mode and SonoAVC (Observer A's ICC 0.980; 95% CI,0.107-0.997; Observer B's ICC,0.971; 95% CI,0.092-0.995). There was good agreement between observers(A and B), the ICC were all 0.999. (2) All data sets could be measured with all three techniques in the group of fetuses.1) In the group of fetuses with normal heart, a high degree of reliability was observed between VOCAL and SonoAVC (left ventricular stroke volume, Observer A's ICC,0.995; 95% CI,0.992-0.997; Observer B's ICC,0.966; 95% CI,0.922-0.982), and between SonoAVC and Inversion (Observer A's ICC,0.977; 95% CI,0.965-0.984; Observer B's ICC,0.970; 95% CI, 0.954-0.980), and between VOCAL and Inversion (Observer A's ICC,0.977; 95% CI, 0.966-0.985; Observer B's ICC,0.981; 95% CI,0.965-0.989). The time necessary to measure the stroke volume was significantly shorter with sonoAVC (3.1 versus 10.8 minutes, and 3.1 versus 6.4 minutes; Both of P<.05) than with VOCAL and Inversion. Bland-Altman tests showed no clinically significant mean percent differences between stroke volume measurements obtained from each ventricle by the same observer or by 2 independent observers using three techniques.2) In the group of fetuses with cardiomegaly, a high degree of reliability was observed between:VOCAL and inversion mode (left ventricular stroke volume, ICC,0.941; 95% CI,0.770-0.985); VOCAL and SonoAVC (ICC,0.900; 95% CI,0.651-0.974); and inversion mode and SonoAVC (ICC,0.935; 95% CI,0.764-0.984). The left ventricular ejection fraction was significantly lower (47.2% versus 66.5%, F=-5.094, P<0.05) than fetuses with normal heart.Conclusions. (1)There was good accuracy of the ultrasound system and measurement techniques applied in volume assessments in the validation experiments. (2)This study shows the feasibility of measuring left ventricular volumes during either systole or diastole and thereby calculating stroke volumes with all three techniques. And there was a high degree of reliability for all three techniques, as shown by excellent ICCs and good intraobserver and interobserver agreement. Thus suggesting that all methods can be used interchangeably. (3)Sonographic automatic volume calculation represents a rapid technique for estimating fetal stroke volume and promises to become the method of choice. (4)There was good accuracy of STIC techniques in measuring fetal left ventricular volumes with cardiomegaly, and the results of this study demonstrated that it would be helpfull for investigating heart function of abnormal fetus.