The purpose of our study was to evaluate the role of contrast-enhanced ultrasound (CEUS) with magnetic resonance imaging (MRI) and computed tomography (CT) in the pathological diagnosis of pancreatic cystic neoplasms (PCNs).A total of 90 patients (66 women, 24 men) aged 18-71 years were studied prospectively. CEUS was performed in all patients, whereas MRI was performed in 85 patients and CT in 69 patients. We analyzed the sensitivity and accuracy of these three imaging modalities to diagnose the PCNs. Neoplasm size, location, shape, intralesional mural nodules, septa and duct dilatation were also assessed by different radiologists.There were no significant differences in sensitivity for discriminating PCNs from pancreatic cystic lesions between CEUS and MRI (p=0.614) or between CEUS and CT (p=0.479). The diagnostic accuracy of CEUS for classifying PCNs was 64.4% (58/90), which was higher than that of CT (53.6%, 37/69, P=0.017), and lower than that of MRI (70.6%, 60/85, p=0.791). Regarding tumor size for lesions larger than 3 cm, CEUS was superior to CT in differentiating the specific type of PCN (p=0.041), and CEUS had the same value as MRI (p=0.774). Furthermore, CEUS is valuable for precisely characterizing internal structures, for instance, septa (p=0.003, compared with CT; p=0.443, compared with MRI) and nodules (p= 0.018, compared with CT; p=0.033, compared with MRI). The number of septa (p=0.033) and cyst morphology (p=0.016) were meaningful indicators in differentiating serous and mucinous adenoma. There was no significant difference in evaluating size and detecting duct dilatation among the three imaging methods.CEUS compares favorably with MRI in displaying the inner structure of PCNs and offers advantages over CT. CEUS can contribute in an important way to the diagnosis of pancreatic cystic neoplasms.© 2020 Sun et al.

Pancreatic cystic lesions are being detected with increasing frequency because of increased use and improved quality of cross-sectional imaging techniques. Pancreatic cystic lesions encompass non-neoplastic lesions (such as pancreatitis-related collections) and neoplastic tumors. Common cystic pancreatic neoplasms include serous cystadenomas, mucinous cystic neoplasms, intraductal papillary mucinous neoplasms, and solid pseudopapillary tumors. These cystic pancreatic neoplasms may have typical morphology, but at times show overlapping imaging features on cross-sectional examinations. This article reviews the classical and atypical imaging features of commonly encountered cystic pancreatic neoplasms and presents the limitations of current cross-sectional imaging techniques in accurately classifying pancreatic cystic lesions.

Mucinous cystic neoplasms (MCN) and serous cystic neoplasms (SCN) account for a large portion of solitary pancreatic cystic neoplasms (PCN). In this study we implemented a convolutional neural network (CNN) model using ResNet50 to differentiate between MCN and SCN. The training data were collected retrospectively from 59 MCN and 49 SCN patients from two different hospitals. Data augmentation was used to enhance the size and quality of training datasets. Fine-tuning training approaches were utilized by adopting the pre-trained model from transfer learning while training selected layers. Testing of the network was conducted by varying the endoscopic ultrasonography (EUS) image sizes and positions to evaluate the network performance for differentiation. The proposed network model achieved up to 0.827 5 accuracy and a 0.88 (95% CI: 0.817–0.930) area under curve (AUC) score. The performance of the implemented deep learning networks in decision-making using only EUS images is comparable to that of traditional manual decision-making using EUS images along with supporting clinical information. Gradient-weighted class activation mapping (Grad-CAM) confirmed that the network model learned the features from the cyst region accurately. This study proves the feasibility of diagnosing MCN and SCN using a deep learning network model. Further improvement using more datasets is needed.

Endoscopic ultrasound (EUS) morphology can aid in the discrimination between mucinous and non-mucinous pancreatic cystic lesions (PCLs) but has several limitations that can be overcome by artificial intelligence. We developed a convolutional neural network (CNN) algorithm for the automatic diagnosis of mucinous PCLs. Images retrieved from videos of EUS examinations for PCL characterization were used for the development, training, and validation of a CNN for mucinous cyst diagnosis. The performance of the CNN was measured calculating the area under the receiving operator characteristic curve (AUC), sensitivity, specificity, and positive and negative predictive values. A total of 5505 images from 28 pancreatic cysts were used (3725 from mucinous lesions and 1780 from non-mucinous cysts). The model had an overall accuracy of 98.5%, sensitivity of 98.3%, specificity of 98.9% and AUC of 1. The image processing speed of the CNN was 7.2 ms per frame. We developed a deep learning algorithm that differentiated mucinous and non-mucinous cysts with high accuracy. The present CNN may constitute an important tool to help risk stratify PCLs.

Clinical histological grading of hepatocellular carcinoma (HCC) differentiation is of great significance in clinical diagnoses, treatments, and prognoses. However, it is challenging for radiologists to evaluate HCC gradings from medical images.In this study, a novel deep neural network was developed by combining the squeeze-and-excitation networks (SENets) in a three-dimensional (3D) densely connected convolutional network (DenseNet), which is referred to as a 3D SE-DenseNet, for the classification of HCC grading using enhanced clinical magnetic resonance (MR) images obtained from two different clinical centers.In the proposed architecture, the SENet was added as an additional layer between the dense blocks of the 3D DenseNet, to mitigate the impact of feature redundancy. For the HCC grading task, the 3D SE-DenseNet was trained after data augmentation, and it outperformed the 3D DenseNet based on the clinical dataset.The quantitative evaluations of the 3D SE-DenseNet on a two-class HCC grading task were conducted based on the dataset, which included 213 samples of the dynamic enhanced MR images. The proposed 3D SE-DenseNet demonstrated an accuracy of 83%, when compared with the 72% accuracy of the 3D DenseNet.Owing to the advantage of useful automatic feature learning by the SE layer, the 3D SE-DenseNet can simultaneously handle useful feature enhancement and superfluous feature suppression. The quantitative experiments confirm the excellent performance of the 3D SE-DenseNet in the evaluation of the HCC grading.Copyright © 2019 Elsevier Ltd. All rights reserved.

The work mainly focused on a validation of the method for determining the content of salicylic acid and individual unknown impurities in new pharmaceutical product—tablets containing: 75, 100 or 150 mg of acetylsalicylic acid and glycine in the amount of 40 mg for each dosage. The separation of the components was carried out by means of HPLC, using a Waters Symmetry C18 column (4.6 × 250 mm, 5 μm) as the stationary phase. The mobile phase consisted of a mixture of 85% orthophosphoric acid, acetonitrile and purified water (2:400:600 V/V/V). Detection was carried out at a wavelength of 237 nm, with a constant flow rate of 1.0 ml min−1. In order to verify the method, linearity, precision (repeatability and reproducibility), accuracy, specificity, range, robustness, system precision, stability of the test and standard solution, limit of quantification and forced degradation were determined. Validation tests were performed in accordance with ICH (International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use) guidelines. The method was validated successfully. It was confirmed that the method in a tested range of 0.005–0.40% salicylic acid with respect to acetylsalicylic acid content is linear, precise and accurate.

Interstitial Lung Disease (ILD) is one of the popular respiratory diseases. The correct diagnosis of ILD is beneficial to improve the effect of treatment for patients. This paper presents an improved DenseNet called small kernel DenseNet (SK-DenseNet) to improve ILD classification performance. According to the characteristics of HRCT features of lung disease, the SK-DenseNet network is more effective to extract high level and small pathological features for ILD classification. Our experiment results show that the proposed SK-DenseNet obtains an outstanding performance (98.4%),which improves 5% performance compared with DenseNet. A comparative analysis with other CNNs, such as AlexNet, VGGNet, ResNet has also demonstrated that the effectiveness of SK-DenseNet in terms of classifying lung disease patterns is superior than those compared ones. The research has validated that using small convolution kernel is useful to improve the recognition efficiency when feature patterns are small.

Transfer learning relaxes the assumption’s limitation requirements of the independent and identical distribution of training data and test data in machine learning. It aims to help the target domain complete the learning task by learning one or more source domains similar to the target domain, solve the problem of scarcity of annotation data and enhance the model’s robustness and generalization performance. The article is a survey on the progress of transfer learning. According to “how to transfer”, transfer learning is divided into four categories: instance-based transfer learning, feature-based transfer learning, model-based transfer learning, and relation-based transfer learning. The paper introduces the basic assumptions, main research questions, common methods, related research of various transfer learning algorithms, and the application of the transfer learning. Finally, we try to point out future research trends.