Search

Number connection test A (NCT-A) and digit symbol test (DST), the preferential neuropsychological tests to detect minimal hepatic encephalopathy (MHE) in China, haven't been standardized in Chinese population. We aimed to establish the norms based on a multi-center cross-sectional study and to detect MHE in cirrhotic patients. NCT-A and DST were administered to 648 healthy controls and 1665 cirrhotic patients. The regression-based procedure was applied to develop demographically adjusted norms for NCT-A and DST based on healthy controls. Age, gender, education, and age by education interaction were all predictors of DST, while age, gender, and education by gender interaction were predictors of log10 NCT-A. The predictive equations for expected scores of NCT-A and DST were established, and Z-scores were calculated. The norm for NCT-A was set as Z ≤ 1.64, while the norm for DST was set as Z ≥ − 1.64. Cirrhotic patients with concurrent abnormal NCT-A and DST results were diagnosed with MHE. The prevalence of MHE was 8.89% in cirrhotic patients, and only worse Child–Pugh classification (P = 0.002, OR = 2.389) was demonstrated to be the risk factor for MHE. The regression-based normative data of NCT-A and DST have been developed to detect MHE in China. A significant proportion of Chinese cirrhotic patients suffered from MHE, especially those with worse Child–Pugh classification. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (DCNNs) have demonstrated promising performance in classifying breast lesions in 2D ultrasound (US) images. Exiting approaches typically use pre-trained models based on architectures designed for natural images with transfer learning. Fewer attempts have been made to design customized architectures specifically for this purpose. This paper presents a comprehensive evaluation on transfer learning based solutions and automatically designed networks, analyzing the accuracy and robustness of different recognition models in three folds. First, we develop six different DCNN models (BNet, GNet, SqNet, DsNet, RsNet, IncReNet) based on transfer learning. Second, we adapt the Bayesian optimization method to optimize a CNN network (BONet) for classifying breast lesions. A retrospective dataset of 3034 US images collected from various hospitals is then used for evaluation. Extensive tests show that the BONet outperforms other models, exhibiting higher accuracy (83.33%), lower generalization gap (1.85%), shorter training time (66 min), and less model complexity (approximately 0.5 million weight parameters). We also compare the diagnostic performance of all models against that by three experienced radiologists. Finally, we explore the use of saliency maps to explain the classification decisions made by different models. Our investigation shows that saliency maps can assist in comprehending the classification decisions. [ABSTRACT FROM AUTHOR]