Automated analysis of fetal cardiac function : a new approach based on tissue Doppler imaging

Abstract: The heart is a central organ in fetal adaptation to various hemodynamic insults and it is possible that subtle changes caused by complications such as hypoxia and anemia could manifest themselves early as altered fetal cardiac function. Color tissue Doppler imaging (cTDI) is an ultrasound technique that evaluates myocardial motion by assessing myocardial velocities. cTDI has been attempted in fetuses, but its use has been hampered by a cumbersome and time-consuming process of image analysis. The main aim of this thesis was to lay the foundation for simplifying the process of image analysis, i.e. the analysis of myocardial velocity traces, using an automated algorithm allowing for quicker and easier assessment of fetal cardiac function during the second half of pregnancy. In all studies, a four-chamber view of the fetal heart was acquired and cine-loops of consecutive cardiac cycles using cTDI were stored before off-line analysis was performed. Myocardial velocity traces describing longitudinal cardiac function in the left (LV) and right ventricular (RV) wall and interventricular septum (IVS) were obtained, and peak myocardial velocities, duration of cardiac cycle time intervals and displacement of the atrioventricular plane assessed. In Study I the feasibility of a beta version of the automated algorithm was evaluated in 261 myocardial velocity traces obtained from 17 echocardiographic examinations in five women. The automated algorithm could analyze 203 out of 261 (78%) myocardial velocity traces. Furthermore, the effect of different sizes of regions of interest (ROIs) on the results at different gestational ages was assessed. With increasing ROI heights there was a loss of velocity information compared to a smaller reference ROI. However, the acceleration traces, used to define cardiac time intervals, were judged to be more well-defined, with increasing ROI heights, in later gestation. In Study II, manual and automated analysis of cTDI velocity traces were compared and the feasibility of using the automated algorithm was assessed in 107 women ≥ 41 weeks of gestation. All myocardial velocity traces (n = 321) were possible to analyze with the manual method of analysis. Myocardial velocities and cardiac cycle time intervals could be measured in 96% of all traces using the automated method of analysis. There were significant positive correlations between all automatically and manually assessed myocardial velocity variables except one. However, only a few of the cardiac cycle time intervals measured by the two techniques correlated significantly, the agreement between methods sometimes showed considerable bias and precision was poor for some cTDI variables. In Study III, gestational age specific reference ranges for normal pregnancies between 18 and 42 weeks of gestation were constructed using the automated analysis of cTDI velocity traces. This was a cross-sectional study with a final study population of 201 pregnant women examined between 18 and 42 weeks of gestation. This study demonstrated an increase of peak myocardial velocities with advancing gestation, whereas the time intervals remained more stable. Study IV included 32 fetuses that underwent 70 intrauterine transfusions (IUTs). Twentyseven fetuses underwent 63 IUTs because of fetal anemia due to maternal alloimmunization and five had other underlying diseases or the anemia was of unknown origin. There were significantly increased myocardial velocities during systole and diastole in the LV wall and IVS, whereas in the RV wall only the systolic velocity was increased before IUT. After IUT, there were significant decreases in the same velocities. When analyzing only first IUTs, there was a significant negative correlation between hemoglobin and myocardial velocity during rapid ventricular filling, i.e. LV Em (rho = -0.61, p = 0.036), as well as the ratio between myocardial velocity during rapid ventricular filling and atrial contraction, i.e. LV Em/Am (rho = -0.82, p = 0.001). This thesis shows that it is possible to use an automated algorithm to analyze cTDI velocity traces to assess fetal cardiac function during the second half of pregnancy. It is proposed that the ROI size should be adjusted according to gestational age as results indicate that this could improve the function of the algorithm. Furthermore, the automated method of analysis was demonstrated feasible at ≥ 41 weeks of gestation. However, sometimes the agreement between the automated and manual methods of analysis of myocardial velocity traces was poor. While none of these methods are considered a gold standard today, the potential benefit of an automated assessment that repeatedly gives the exact same results is an important prerequisite for the future application and use of cTDI in research and clinical practice. Gestational age specific reference ranges in the second half of normal pregnancy were constructed, which is also a pre-requisite to discriminate normal from abnormal cardiac function. Finally, the automated analysis was evaluated in fetuses undergoing IUT due to suspected fetal anemia, and illustrated increased myocardial velocities before transfusion that decreased with reversal and cessation of the hyperdynamic state after transfusion in accordance with previous findings in fetal blood flow velocities. To conclude, cTDI with automated analysis shows promise as a feasible method to study fetal cardiac function and to detect cardiac dysfunction and adaptation to pathological situations. The use of an automated analysis could also facilitate its application in research and clinical practice.