29th World Congress on Ultrasound in Obstetrics and Gynecology
Objectives: Determine the accuracy of detection and diagnosis in
the first trimester of congenital heart defects (CHD) at our centre.
Methods: This is a retrospective descriptive study which analyses all
CHD detected in the first trimester of pregnancy at our Prenatal Diag-
nosis Unit. Inclusion criteria were all fetuses with CRL between 45
and 84 mm diagnosed with a CHD between January 2014 and March
2019. Diagnostic accuracy of CHD detection in the first trimester
was analysed by comparing it with postnatal diagnosis or necropsy.
Results: A total of 71 fetuses with CHD were detected in the
first trimester. The detection rate was 100%. In 82%, an accurate
diagnosis of the CHD was made, while in 13 cases precise diagnosis
could not be established in the first trimester. In seven of these 13
cases a conotruncal anomaly was suspected and confirmed in the
2nd trimester, where the exact type of conotruncal anomaly was
determined. In the other 6 cases, a complex CHD was suspected
without being able to classify it in the 1st trimester of pregnancy
confirming the CHD in the 2nd trimester. Invasive procedures were
performed in 87.3%, performing chorionic villus sampling in 91.9%
of cases. The karyotype was normal in 23 patients (34.3%) and
65.7% presented chromosome abnormalities. 54 patients (76.1%)
terminated their pregnancy. In all cases the diagnosis was confirmed
postnatally or by necropsy.
Conclusions: Accurate diagnosis of CHD in the first trimester,
combining the transabdominal and transvaginal routes, is possible
in the majority of cases. The CHD that presents highest diagnostic
difficulty in the 1st trimester are the conotruncal anomalies, where
the defect can be suspected, though it may be difficult to differentiate
the different types of conotruncal anomalies.
Supporting information can be found in the online
version of this abstract
OP15.03
Prenatal and newborn screening for congenital heart disease
L. Cao1 , Y. Ren2
1
Ultrasonic Diagnosis Department of Obstetrics and Gynecol-
ogy Hospital, Fudan University, Shanghai, China, Shanghai,
China; 2 Ultrasound Department, Shanghai, Obstetrics and
Gynecology Hospitai of Fudan Uinversity, Shanghai, China
Objectives: To investigate the value of prenatal and postnatal
echocardiogram in the diagnosis of congenital heart disease.
Methods: Pregnant women who were born in the hospital
from 2017.1 to 2017.12 were included in the study. The fetal
structure was routinely screened during the second trimester. The
newborns were screened for congenital heart disease by pulse
oximetry and auscultation. The positive ones were examined by
neonatal echocardiography. At 42 days postpartum, the returning
infants underwent routine auscultationandthose who were examined
positive were further examined by echocardiography.
Results: A total of 13,849 pregnant women were collected during
the study. 62 cases of congenital heart disease were found before
birth, including 34 cases of severe congenital heart disease and 28
cases of mild congenital heart disease. In the neonatal period, 192
cases of congenital heart disease were found, including 6 cases of
severe congenital heart disease and 186 cases of mild congenital heart
disease. The incidence of congenital heart disease is 18.3%. The most
common severe congenital heart disease is tetralogy of Fallot, a total
of 17 cases. The cases of prenatal missed diagnosis were 3 cases
of tetralogy of Fallot, 2 cases of total abnormal pulmonary venous
drainage and 1 case of transposition of great arteries. The more
common congenital heart diseases in the neonatal period were atrial
septal defect and ventricular septal defect.
134
Short oral presentation abstracts
Conclusions: Fetal echocardiography is more sensitive to the exam-
ination of severe congenital heart disease, and postpartum neonatal
echocardiography can detect more mild congenital heart disease.
OP15.04
Novel AI-guided ultrasound screening system for fetal heart
can demonstrate findings in timeline diagram
R. Komatsu1 , R. Matsuoka1 , T. Arakaki1 , M. Tokunaka1 ,
M. Komatsu2,3 , A. Sakai4,5 , S. Yasutomi4,6 , R. Hamamoto2,3 ,
A. Sekizawa1
1
Obstetrics and Gynecology, Showa University School of
Medicine, Tokyo, Japan; 2 AIP Center, Riken, Tokyo, Japan;
3 Division of Molecular Modification and Cancer Biology,
National Cancer Center Research Institute, Tokyo, Japan;
4
Riken AIP-Fujitsu Collaboration Center, Riken, Tokyo,
Japan; 5 AI Service Business Unit, Fujitsu Ltd, Tokyo, Japan;
6
Artificial Intelligence Laboratory, Fujitsu Laboratories Ltd,
Kawasaki, Japan
Objectives: We have developed a novel fetal cardiac screening
system that is used deep leaning technology. This system
can real-timely and automatically detect each part in fetal
echocardiographic movie that was acquired in 2nd trimester cardiac
screening examination. Artificial intelligence (AI) was based on
convolutional neural networks. Whereas, the decision process of AI,
especially deep learning, has been unexplainable and still in black
box. Therefore, we tried to visualise and explain its process toward
practical utilisation of our system in clinic.
Methods: 42 normal fetal heart movies were used as a training
dataset, and the experts had annotated the correct positions of 18
different parts of the heart and peripheral organs such as the atrium,
ventricle, blood vessels and stomach. The fetal movie of tetralogy
of Fallot (TOF) and transposition of the great arteries (TGA) were
used as a test dataset. Each movie includes sequential axial views
from the level of stomach to base of heart. AI described each part as
barcode-used diagram along with the sweep scanning timeline. The
diagram was demonstrated the detection probability of each part in
each moment.
Results: The diagram clearly showed the abdomen part, heart
structural part, and out flow tract/blood vessel part as like we
observed in the sweep movie. In case of TOF, the pulmonary artery
was not demonstrated clearly and in case of TGA, the detection
pattern of out flow tract/blood vessel part was different from the
normal fetal ones.
Conclusions: We have successfully developed the diagram system
demonstrating the object detection probability to visualise how AI
detect and demonstrate fetal cardiac parts in ultrasound screening.
Furthermore, this barcode-used diagram should enable to list
the differential diagnoses and be informative to lead the expert
examination.
Supporting information can be found in the online
version of this abstract
OP15.05
Risk of spontaneous and iatrogenic preterm birth in
pregnancies with prenatal diagnosis of major congenital heart
disease
V. Giorgione1,2 , V. Fesslova3 , S. Boveri3 , M. Candiani4 ,
A. Khalil1,2 , P.I. Cavoretto4
1
Fetal Medicine Unit, Department of Obstetrics and Gynae-
cology, St George’s University Hospitals NHS Foundation
Trust, London, United Kingdom; 2 Vascular Biology Research
© The Authors 2019
© Ultrasound in Obstetrics & Gynecology 2019; 54 (Suppl. 1): 84–154.