It is still challenging to make accurate diagnosis of biliary atresia (BA) with sonographic gallbladder images particularly in rural area without relevant expertise. To help diagnose BA based on sonographic gallbladder images, an ensembled deep learning model is developed. The model yields a patient-level sensitivity 93.1% and specificity 93.9% [with areas under the receiver operating characteristic curve of 0.956 (95% confidence interval: 0.928-0.977)] on the multi-center external validation dataset, superior to that of human experts. With the help of the model, the performances of human experts with various levels are improved. Moreover, the diagnosis based on smartphone photos of sonographic gallbladder images through a smartphone app and based on video sequences by the model still yields expert-level performances. The ensembled deep learning model in this study provides a solution to help radiologists improve the diagnosis of BA in various clinical application scenarios, particularly in rural and undeveloped regions with limited expertise.

Early diagnosis of congenital heart disease (CHD) can improve the prognosis of neonates with CHD. We retrospectively evaluated the value of prenatal diagnosis of CHD by comparing the pregnancy outcomes. Prenatal diagnosis of CHD was established by echocardiographic evaluation of fetal heart. Amniotic fluid and/or cord blood genetic examination, pathological anatomy, casting specimen, and/ or multidisciplinary-joint consultation (MDJC) were performed. A total of 1492 fetuses with CHD were diagnosed by prenatal echocardiography from 67834 pregnant women. There were 445, 236, 583, and 228 cases in groups A (simple CHD), B (simple CHD plus extra-cardiac abnormality), C (complex CHD), and D (complex CHD plus extra-cardiac abnormality), respectively. The pregnancy continuation rate in the four groups was 98.67%, 85.71%, 67.65%, and 36.84%, respectively (P < 0.001). The pregnancy termination rate for fetal CHD with extra-cardiac abnormalities was significantly higher than that for fetuses with only CHD (81.24% vs. 53.6%, P < 0.05). Prenatal genetic test revealed chromosomal abnormalities in 20.43% of fetuses with CHD. MDJC significantly decreased the pregnancy termination rate. In 88 cases, the original decision to terminate the pregnancy was changed after consultation and the pregnancy was continued. Of these, 87 cases culminated in live births; 65 of these children had better prognosis. Nine-segment sequential segment analysis method for prenatal fetal echocardiography was compared with the results of pathological anatomy, cast specimen, postoperative diagnosis, and postnatal ultrasound. The accuracy of prenatal ultrasound for diagnosis of fetal complex CHD and fetal simple CHD was 90.5-91.66% and 98.6%, respectively. Prenatal ultrasound is still the most effective method for fetal CHD diagnosis. According to the "China Birth Defect Prevention Report (2012)" 1 , congenital heart disease (CHD) is the most common type of birth defect in China accounting for 26.7% of all birth defects in 2011. More than 130,000 new cases of CHD are reported each year in China. Globally, a total of 48.9 million cases of CHD were reported in 2015 2. The reported incidence of CHD varies from 4 to 75 per 1,000 live births depending upon the method of diagnosis. About 0.6 to 1.9% of live born infants with CHD have moderate to severe disease 3. CHD is the leading cause of birth defect-related deaths 4. In 2015, an estimated 303,300 infants died around the world due to CHD 5 .

Abnormal origin of the pulmonary artery has a characteristic ultrasonographic appearance. The branch of the pulmonary artery cross section can provide important clues to the diagnosis of abnormal pulmonary artery origin. Pulmonary artery abnormalities are often associated with pulmonary atresia or stenosis.

No abstract

Objectives To develop and validate a biliary atresia (BA) diagnostic score based on serum gamma‐glutamyl transferase (GGT) levels and conventional ultrasound features for discriminating BA in patients with jaundice from two centers. Methods A total of 958 patients from one hospital were classified as the derivation cohort, and 725 patients from another hospital were classified as the validation cohort. The optimal GGT cutoff value for diagnosing BA was calculated in the derivation cohort and subsequently verified in the validation cohort. Gallbladder abnormalities and the triangular cord (TC) sign were evaluated in all patients. A BA diagnostic score was developed for diagnosing BA using the GGT levels, gallbladder abnormalities and the TC sign based on the data from the derivation cohort followed by external validation. Results Based on the optimal cutoff value 350.0 U/L, GGT yielded a sensitivity of 59.3% and specificity of 85.4% in diagnosing BA. The area under the receiver operating characteristic curve (AUC 0.724) was inferior to that of the gallbladder (AUC 0.911, P < .001) and comparable to that of the TC sign (AUC 0.771, P = .128). The combination of GGT and ultrasound diagnosis could help to reduce the misdiagnosis of 9 infants with BA. The BA diagnostic score yielded a sensitivity of 93.3% and specificity of 95.0% with the highest AUC in this study (0.941). Conclusions GGT can add diagnostic value to ultrasound examination when diagnosing BA. The BA diagnostic score based on GGT, gallbladder abnormalities and the TC sign shows satisfactory discrimination abilities in BA.