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.
Purpose. To investigate the diagnostic performance of shear wave elastography (SWE) for measuring liver stiffness to identify and differentiate biliary atresia (BA) from cholestatic hepatitis in infants younger than 90 days. Methods. A total of 138 infants younger than 90 days with cholestatic hepatitis were examined by SWE. The infants were subclassified into BA and nonbiliary atresia (non-BA) groups. Receiver operating characteristic (ROC) analysis was used to determine the sensitivity and specificity of hepatic Young’s modulus measurements, the ultrasonic findings in the differential diagnosis of suspected BA, and the cut-off value to diagnose BA. Results. In all infants with cholestatic hepatitis, the cut-off value of hepatic Young’s modulus to differentiate the BA group from the non-BA group was 12.35 kPa and the area under the ROC curve (AUC) was 0.937, with a sensitivity of 84.3% and a specificity of 89.7%; nevertheless the AUC of the abnormal gallbladder (AbGB) was 0.940, with a sensitivity of 96.1% and a specificity of 92.0%. In the parallel test, triangular cord (TC) sign combined with AbGB had the best diagnostic performance and the AUC was 0.960, with a sensitivity of 100% and a specificity of 92.0%. In the serial test, SWE combined with AbGB achieved the best diagnostic performance; the AUC was 0.902, the sensitivity and specificity were 80.4% and 100%, respectively. Conclusions. SWE could not only help differentiate BA from cholestatic hepatic diseases but also increase the diagnostic specificity when combined with grey-scale ultrasound in the serial test.
Timely and accurate diagnosis of severe neonatal cerebral lesions is critical for preventing long-term neurological damage and addressing life-threatening conditions. Cranial ultrasound is the primary screening tool, but the process is time-consuming and reliant on operator’s proficiency. In this study, a deep-learning powered neonatal cerebral lesions screening system capable of automatically extracting standard views from cranial ultrasound videos and identifying cases with severe cerebral lesions is developed based on 8,757 neonatal cranial ultrasound images. The system demonstrates an area under the curve of 0.982 and 0.944, with sensitivities of 0.875 and 0.962 on internal and external video datasets, respectively. Furthermore, the system outperforms junior radiologists and performs on par with mid-level radiologists, with 55.11% faster examination efficiency. In conclusion, the developed system can automatically extract standard views and make correct diagnosis with efficiency from cranial ultrasound videos and might be useful to deploy in multiple application scenarios.