A multicenter study on two-stage transfer learning model for duct-dependent CHDs screening in fetal echocardiography.

Duct-dependent congenital heart diseases (CHDs) are a serious form of CHD with a low detection rate, especially in underdeveloped countries and areas. Although existing studies have developed models for fetal heart structure identification, there is a lack of comprehensive evaluation of the long axis of the aorta. In this study, a total of 6698 images and 48 videos are collected to develop and test a two-stage deep transfer learning model named DDCHD-DenseNet for screening critical duct-dependent CHDs. The model achieves a sensitivity of 0.973, 0.843, 0.769, and 0.759, and a specificity of 0.985, 0.967, 0.956, and 0.759, respectively, on the four multicenter test sets. It is expected to be employed as a potential automatic screening tool for hierarchical care and computer-aided diagnosis. Our two-stage strategy effectively improves the robustness of the model and can be extended to screen for other fetal heart development defects.

Fetal cardiac geometry and function in pregnancies with well-controlled gestational diabetes mellitus using Fetal HQ.

OBJECTIVE: To determine whether abnormal cardiac shape and ventricular global, transverse, and longitudinal contractility are present in fetuses of women with well-controlled GDM. METHODS: A prospective observational study was performed on 80 fetuses of women with well-controlled GDM and 90 control fetuses. Using Fetal HQ, a new speckle-tracking technique, cardiac shape, global contractility, transverse contractility, and longitudinal contractility were calculated. The number and percentage of fetuses with z score values below 5th or above 95th were computed. RESULTS: Compared with controls, there were no significant differences in the frequency of cardiac geometric abnormalities in GDM fetuses. Despite good glycemic control, 60.0% of fetuses in the well-controlled GDM group had one or more types of global, longitudinal, and transverse contractility abnormalities of one or both ventricles, but more frequent on the right ventricle (RV, 50%). The most frequent abnormality of the RV occurred in the transverse contractility (35%), followed by abnormalities of global contractility (25%), and longitudinal contractility (21.3%), compared with controls. The left ventricle (LV) analysis demonstrated that the percentage of study fetuses with only transverse contractility abnormality (18.8%) was significantly higher. CONCLUSIONS: Despite good glycemic control, abnormal ventricular contractility was present in fetuses of women with GDM, but more frequent in the RV. For both the RV and LV, transverse ventricular contractility abnormality were more prevalent than abnormal global and longitudinal contractility. Fetuses of women with GDM should be evaluated for ventricular contractility abnormality and have more follow-ups despite good glycemic control.

Echocardiographic findings for improved prenatal diagnosis of aortic coarctation with ventricular septal defect.

Accurate prenatal diagnosis of coarctation of the aorta (CoA) associated with ventricular septal defect (VSD) remains challenging. The objective of the study was to identify which Doppler and/or two-dimensional sonographic findings are most useful for predicting fetal CoA/VSD. A retrospective cohort study identified 35 fetuses with suspected CoA/VSD. Prenatal imaging characteristics included the right ventricular/left ventricular, pulmonary artery (PA)/aorta ratio, aortic isthmus (AOI) Z score, diastolic velocity-time integral (VTID), and systolic velocity-time integral (VTIS) at the AOI. The area under the receiver operating characteristic curve (AUC), integrated discrimination improvement (IDI), and net reclassification improvement (NRI) were calculated. Significant differences in the PA/AO, VTID, VTID/VTIS, VTID/VTIS, VTID/(VTID + VTIS), and AOI Z score between the true CoA group and false positives were found. When associated with VSD, the VTID/VTIS and VTID/(VTID + VTIS) had the highest AUC (0.97, 95% confidence interval: 0.84-1.00), with 88.46% sensitivity and 100.00% specificity for predicting the true CoA. The AOI Z score had the highest sensitivity (92.31%). Adding the VTID/VTIS to the AOI Z score significantly improved the performance (IDI, 50%; NRI, 82%; P < 0.05), with an improvement in specificity (77.78% vs. 55.56%; non-Event P = 0.008) without sacrificing sensitivity (96.15% vs. 92.31%; Event P = 0.564). In fetuses with suspected CoA associated with VSD, the quantitative spectral Doppler metric aided accurate detection of the fetal CoA, with reduced false positives. The conventional AOI Z score plus spectral Doppler metric may improve the overall diagnostic accuracy of CoA/VSD.