Objective To achieve automatic segmentation of cell nuclei in gastrointestinal cancer pathological images by using a deep learning algorithm,so as to assist in the quantitative analysis of subsequent pathological images.Methods A total of 59 patients with gastrointestinal cancer treated in Ruijin Hospital,Shanghai Jiao Tong University School of Medicine from Jan 2022 to Feb 2022,were selected as the research objects.Python and LabelMe were used for data anonymization,image segmentation,and region of interest annotation of patients'pathological images.A total of 944 pathological images were included,and 9 703 nuclei were annotated.Then,a new semantic segmentation model based on deep learning was constructed.The model introduced deformable attention transformer(DAT)to realize automatic,accurate and efficient segmentation of pathological image nuclei.Finally,multiple segmentation evaluation criteria are used to evaluate the model's performance.Results The mean absolute error of the segmentation results of the model proposed in this paper was 0.112 6,and the dice coefficient(Dice)was 0.721 5.Its effect was significantly better than the U-net baseline model,and it was ahead of models such as ResU-net++,R2Unet and R2AttUnet.Moreover,the segmentation results were relatively stable with good generalization.Conclusion The segmentation model established in this study can accurately identify and segment the nuclei in the pathological images,with good robustness and generalization,which is helpful to play an auxiliary diagnostic role in practical applications.

Objective: To identify the risk factors in mortality of pediatric acute respiratory distress syndrome (PARDS) in pediatric intensive care unit (PICU). Methods: Second analysis of the data collected in the "efficacy of pulmonary surfactant (PS) in the treatment of children with moderate to severe PARDS" program. Retrospective case summary of the risk factors of mortality of children with moderate to severe PARDS who admitted in 14 participating tertiary PICU between December 2016 to December 2021. Differences in general condition, underlying diseases, oxygenation index, and mechanical ventilation were compared after the group was divided by survival at PICU discharge. When comparing between groups, the Mann-Whitney U test was used for measurement data, and the chi-square test was used for counting data. Receiver Operating Characteristic (ROC) curves were used to assess the accuracy of oxygen index (OI) in predicting mortality. Multivariate Logistic regression analysis was used to identify the risk factors for mortality. Results: Among 101 children with moderate to severe PARDS, 63 (62.4%) were males, 38 (37.6%) were females, aged (12±8) months. There were 23 cases in the non-survival group and 78 cases in the survival group. The combined rates of underlying diseases (52.2% (12/23) vs. 29.5% (23/78), χ²=4.04, P=0.045) and immune deficiency (30.4% (7/23) vs. 11.5% (9/78), χ²=4.76, P=0.029) in non-survival patients were significantly higher than those in survival patients, while the use of pulmonary surfactant (PS) was significantly lower (8.7% (2/23) vs. 41.0% (32/78), χ²=8.31, P=0.004). No significant differences existed in age, sex, pediatric critical illness score, etiology of PARDS, mechanical ventilation mode and fluid balance within 72 h (all P>0.05). OI on the first day (11.9(8.3, 17.1) vs.15.5(11.7, 23.0)), the second day (10.1(7.6, 16.6) vs.14.8(9.3, 26.2)) and the third day (9.2(6.6, 16.6) vs. 16.7(11.2, 31.4)) after PARDS identified were all higher in non-survival group compared to survival group (Z=-2.70, -2.52, -3.79 respectively, all P<0.05), and the improvement of OI in non-survival group was worse (0.03(-0.32, 0.31) vs. 0.32(-0.02, 0.56), Z=-2.49, P=0.013). ROC curve analysis showed that the OI on the thind day was more appropriate in predicting in-hospital mortality (area under the curve= 0.76, standard error 0.05,95%CI 0.65-0.87,P<0.001). When OI was set at 11.1, the sensitivity was 78.3% (95%CI 58.1%-90.3%), and the specificity was 60.3% (95%CI 49.2%-70.4%). Multivariate Logistic regression analysis showed that after adjusting for age, sex, pediatric critical illness score and fluid load within 72 h, no use of PS (OR=11.26, 95%CI 2.19-57.95, P=0.004), OI value on the third day (OR=7.93, 95%CI 1.51-41.69, P=0.014), and companied with immunodeficiency (OR=4.72, 95%CI 1.17-19.02, P=0.029) were independent risk factors for mortality in children with PARDS. Conclusions: The mortality of patients with moderate to severe PARDS is high, and immunodeficiency, no use of PS and OI on the third day after PARDS identified are the independent risk factors related to mortality. The OI on the third day after PARDS identified could be used to predict mortality.
Female ; Male ; Humans ; Child, Preschool ; Infant ; Child ; Critical Illness ; Pulmonary Surfactants/therapeutic use* ; Retrospective Studies ; Risk Factors ; Respiratory Distress Syndrome/therapy*

Objective: To establish a comprehensive diagnostic classification model of lateral cephalograms based on artificial intelligence (AI) to provide reference for orthodontic diagnosis. Methods: A total of 2 894 lateral cephalograms were collected in Department of Orthodontics, Capital Medical University School of Stomatology from January 2015 to December 2021 to construct a data set, including 1 351 males and 1 543 females with a mean age of (26.4± 7.4) years. Firstly, 2 orthodontists (with 5 and 8 years of orthodontic experience, respectively) performed manual annotation and calculated measurement for primary classification, and then 2 senior orthodontists (with more than 20 years of orthodontic experience) verified the 8 diagnostic classifications including skeletal and dental indices. The data were randomly divided into training, validation, and test sets in the ratio of 7∶2∶1. The open source DenseNet121 was used to construct the model. The performance of the model was evaluated by classification accuracy, precision rate, sensitivity, specificity and area under the curve (AUC). Visualization of model regions of interest through class activation heatmaps. Results: The automatic classification model of lateral cephalograms was successfully established. It took 0.012 s on average to make 8 diagnoses on a lateral cephalogram. The accuracy of 5 classifications was 80%-90%, including sagittal and vertical skeletal facial pattern, mandibular growth, inclination of upper incisors, and protrusion of lower incisors. The acuracy rate of 3 classifications was 70%-80%, including maxillary growth, inclination of lower incisors and protrusion of upper incisors. The average AUC of each classification was ≥0.90. The class activation heat map of successfully classified lateral cephalograms showed that the AI model activation regions were distributed in the relevant structural regions. Conclusions: In this study, an automatic classification model for lateral cephalograms was established based on the DenseNet121 to achieve rapid classification of eight commonly used clinical diagnostic items.
Male ; Female ; Humans ; Young Adult ; Adult ; Artificial Intelligence ; Deep Learning ; Cephalometry ; Maxilla ; Mandible/diagnostic imaging*

Objective: To develop a multi-classification orthodontic image recognition system using the SqueezeNet deep learning model for automatic classification of orthodontic image data. Methods: A total of 35 000 clinical orthodontic images were collected in the Department of Orthodontics, Capital Medical University School of Stomatology, from October to November 2020 and June to July 2021. The images were from 490 orthodontic patients with a male-to-female ratio of 49∶51 and the age range of 4 to 45 years. After data cleaning based on inclusion and exclusion criteria, the final image dataset included 17 453 face images (frontal, smiling, 90° right, 90° left, 45° right, and 45° left), 8 026 intraoral images [frontal occlusion, right occlusion, left occlusion, upper occlusal view (original and flipped), lower occlusal view (original and flipped) and coverage of occlusal relationship], 4 115 X-ray images [lateral skull X-ray from the left side, lateral skull X-ray from the right side, frontal skull X-ray, cone-beam CT (CBCT), and wrist bone X-ray] and 684 other non-orthodontic images. A labeling team composed of orthodontic doctoral students, associate professors, and professors used image labeling tools to classify the orthodontic images into 20 categories, including 6 face image categories, 8 intraoral image categories, 5 X-ray image categories, and other images. The data for each label were randomly divided into training, validation, and testing sets in an 8∶1∶1 ratio using the random function in the Python programming language. The improved SqueezeNet deep learning model was used for training, and 13 000 natural images from the ImageNet open-source dataset were used as additional non-orthodontic images for algorithm optimization of anomaly data processing. A multi-classification orthodontic image recognition system based on deep learning models was constructed. The accuracy of the orthodontic image classification was evaluated using precision, recall, F1 score, and confusion matrix based on the prediction results of the test set. The reliability of the model's image classification judgment logic was verified using the gradient-weighted class activation mapping (Grad-CAM) method to generate heat maps. Results: After data cleaning and labeling, a total of 30 278 orthodontic images were included in the dataset. The test set classification results showed that the precision, recall, and F1 scores of most classification labels were 100%, with only 5 misclassified images out of 3 047, resulting in a system accuracy of 99.84%(3 042/3 047). The precision of anomaly data processing was 100% (10 500/10 500). The heat map showed that the judgment basis of the SqueezeNet deep learning model in the image classification process was basically consistent with that of humans. Conclusions: This study developed a multi-classification orthodontic image recognition system for automatic classification of 20 types of orthodontic images based on the improved SqueezeNet deep learning model. The system exhibitted good accuracy in orthodontic image classification.
Humans ; Male ; Female ; Child, Preschool ; Child ; Adolescent ; Young Adult ; Adult ; Middle Aged ; Deep Learning ; Reproducibility of Results ; Radiography ; Algorithms ; Cone-Beam Computed Tomography

Objective: To investigate the predictive value of SYNTAX-Ⅱ score on long term prognosis of patients diagnosed with chronic total occlusion (CTO) and received percutaneous coronary intervention (PCI). Methods: Patients undergoing CTO-PCI in Fuwai hospital from January 2010 to December 2013 were enrolled in this retrospective analysis. The SYNTAX-Ⅱ score of the patients was calculated. According to SYNTAX-Ⅱ score tertiles, patients were stratified as follows: SYNTAX-Ⅱ≤20, 2027. Primary endpoint was major adverse cardiac events (MACCE), including all-cause death, myocardial infarction, stroke and any revascularization. Secondary endpoints included stent thrombosis, heart failure and target lesion failure (TLF). Patients were followed up by outpatient visit or telephone call at 1 month, 6 months and 1 year after PCI, and annually up to 5 years. Multivariate Cox regression model was used to analyze the independent risk factors of all-cause death in patients undergoing CTO-PCI. The predictive value of SYNTAX score with SYNTAX-Ⅱ score for all-cause death was evaluated by the receiver operating characteristic (ROC) curve and the area under the curve (AUC). Results: A total of 2 391 patients with CTO and received PCI were enrolled in this study. The mean age was (57.0±10.5) years, 1 994 (83.40%) patients were male. There were 802 patients in lower tertile group (SYNTAX-Ⅱ≤20), 798 patients in intermediate group (2027). At the end of 5-year follow-up, the loss to follow-up rate of the three groups was 9.10%(73/802), 10.78%(86/798)and 8.85%(70/791), respectively. The rate of all-cause mortality (1.78% (13/729) vs. 3.65% (26/712) vs. 9.02% (65/721), P<0.001), cardiac death (1.37% (10/729) vs. 2.11% (15/712) vs. 4.85% (35/721), P<0.001), target vessel myocardial infarctions (4.25% (31/729) vs. 4.49% (32/712) vs. 7.07% (51/721), P=0.03), probable stent thrombosis (1.51% (11/729) vs. 2.81% (20/712) vs. 3.61% (26/721), P=0.04) and heart failure (1.78% (13/729) vs. 1.97% (14/712) vs. 5.41% (39/721), P<0.001) increased in proportion to increasing SYNTAX-Ⅱ score (all P<0.05). Multivariable Cox regression analysis indicated that female (HR=2.05, 95%CI 1.12-3.73, P=0.01), left ventricular ejection fraction (HR=0.97, 95%CI 0.95-1.00, P=0.05) and SYNTAX-Ⅱ score (HR=1.07, 95%CI 1.02-1.11,P=0.01) were independent predictors for all-cause mortality in patients undergoing CTO-PCI. The predicted value of the SYNTAX-Ⅱ score for all-cause death was significantly higher than the SYNTAX score (AUC 0.71 vs. 0.60, P=0.003). Conclusion: For CTO patients who underwent percutaneous coronary intervention, SYNTAX-Ⅱ score is an independent predictor for 5-year all-cause death, and SYNTAX-Ⅱ serves as an important predictor for all-cause death in these patients.
Humans ; Male ; Female ; Middle Aged ; Aged ; Percutaneous Coronary Intervention ; Coronary Artery Disease ; Retrospective Studies ; Stroke Volume ; Treatment Outcome ; Ventricular Function, Left ; Myocardial Infarction ; Prognosis ; Risk Factors ; Heart Failure ; Coronary Occlusion/surgery*

Objective: To summarize and analyze the clinical characteristics and gene mutations of 6 patients with Wiedemann-Steiner syndrome (WDSTS). Methods: To review and analyze the clinical data, including general conditions, clinical manifestations, growth hormone, cranial or pituitary gland magnetic resonance imaging (MRI),gene results and other data, 6 cases with WDSTS admitted to the Department of Endocrinology, Genetics and Metabolism of Jiangxi Provincial Children's Hospital and the Department of Child Care of Pingxiang Maternity and Child Care from April 2017 to February 2021 were recruited. Results: Of the 6 patients, 2 were male and 4 were female. The age of the first visit ranged from 1.0 to 11.2 years. All the 6 children presented with growth retardation and mental retardation and they all had typical facial dysmorphism and hypertrichosis (mainly on the back and limbs). Among them, case 5 had a growth hormone deficiency, and case 2 and 4 had abnormalities revealed by cranial MRI. Variations in KMT2A gene were identified in these 6 patients: c.10900+2T>C,c.10837C>T(p.Gln3613*), c.4332G>A(p.E1444E), c.2508dupC(p.W838Lfs*9), c.11695_11696delinsT(p.T3899Sfs*73), c.9915dupA (p.P3306Tfs*22).Among these variations, c.4332G>A, c.11695_11696delinsT and c.9915dupA were novel mutations. Therefore, the final diagnosis of these patients was WDSTS. Conclusions: Patients presented with short stature and mental retardation, typical facial dysmorphism and hypertrichosis should be considered WDSTS. Whole-exome sequencing plays an important role in disease diagnosis and genetic counseling.
Abnormalities, Multiple ; Child ; Child, Preschool ; Craniofacial Abnormalities ; Female ; Growth Disorders/genetics* ; Histone-Lysine N-Methyltransferase ; Humans ; Hypertrichosis/genetics* ; Infant ; Intellectual Disability/genetics* ; Male ; Myeloid-Lymphoid Leukemia Protein ; Pregnancy ; Syndrome

ObjectiveTo investigate the therapeutic effect of Xuebijing injection （XBJ） on sodium taurocholate （Na-Tc）-induced severe acute pancreatitis （SAP） in rats. MethodForty rats were randomly assigned into 5 groups： sham operation group， SAP model group， and low-， medium-， and high-dose （4， 8， 12 mL·kg·d^-1， respectively） XBJ groups. SAP model was established by retrograde injection of Na-Tc （1 mL·kg^-1） into the biliary and pancreatic ducts. XBJ was injected intraperitoneally 3 days before and 0.5 h after modeling. The ascitic fluid volume and the pancreas weight-to-body weight ratio were measured. The pathological changes of pancreatic tissue were observed via hematoxylin-eosin （HE） staining. The protein levels of formyl peptide receptor 1 （FPR1） and nucleotide-binding oligomerization domain-like receptor 3 （NLRP3） in pancreatic tissue were detected by immunohistochemistry. Western blot was employed to determine the expression levels of NADH-ubiquinone oxidoreductase chains 1-6 （MT-ND1， MT-ND2， MT-ND3， MT-ND4， MT-ND5， and MT-ND6） in rat plasma. ResultCompared with sham operation group， the SAP model group showcased increased ascitic fluid volume and pancreas weight-to-body weight ratio （P<0.05）， serious lesions in pancreatic tissue， increased total pathological score （P<0.05）， and up-regulated protein levels of FPR1 and NLRP3 in pancreatic tissue （P<0.05）. The model group had lower MT-ND2 level （P<0.05） and higher MT-ND1， MT-ND3， and MT-ND6 levels in plasma （P<0.05） than the sham operation group， while MT-ND4 and MT-ND5 had no significant differences between the two groups. Compared with SAP model group， the XBJ treatment decreased ascitic fluid volume and pancreas weight-to-body weight ratio （P<0.01）， ameliorated pancreatic lesions， and down-regulated the protein levels of FPR1 and NLRP3 in pancreatic tissue （P<0.01）. The treatments， especially high-dose XBJ （P<0.01）， down-regulated the expression of MT-ND1 （P<0.01）， MT-ND3 （P<0.01）， MT-ND6 （P<0.01）， and MT-ND4 and did not change that of MT-ND5. ConclusionXBJ may antagonize partial mitochondrial N-formyl peptides and excessive inflammatory response mediated by FPR1/NLRP3 to treat SAP in rats.

One undescribed diterpenoid acid and six compounds were isolated from the 95% ethanol fraction of Pinus kesiya var. langbianensis (A.Chev.) Gaussen ex Bui resin by using various chromatographic methods, including MCI Gel, Sephadex LH-20, ODS, silica gel and semi-preparative HPLC. The planar structures were identified by spectroscopy methods (1D, 2D NMR, UV, IR, MS, etc.), and the absolute configuration of the new compound was determined by ECD calculation. Compound 1 is a new compound, and compounds 2, 5-7 were isolated from Pinus kesiya var. langbianensis (A.Chev.) Gaussen ex Bui for the first time.

This study investigated the effects of ginkgolide B on the long-chain fatty acid metabolism-related enzyme protein peroxisome proliferators-activated receptors α (PPARα), long-chain specific acyl-CoA dehydrogenase (LCAD), carnitine palmitoyl transterase-1 (CPT-1), and acyl coenzyme A oxidase 1 (ACOX1) expression in the liver of rats with non-alcoholic fatty liver disease (NAFLD). All the animal welfare and experimental procedures are in accordance with the regulations of the Animal Ethics Committee of Yunnan University of Traditional Chinese Medicine. After successfully building the rat model of non-alcoholic abnormal liver disease, the rats were divided into the model group, the simvastatin group, and the low-dose, middle-dose, and high-dose groups of ginkgolide B according to random number method, and were given corresponding drug treatment 4 weeks. We detected liver pathological indicators and determined blood lipids, transaminase and anti-oxidation indexes. Western blot and RT-PCR assays were used to detect the protein and mRNA levels of PPARα, LCAD, CPT-1, and ACOX1 in livers. The results showed that: ① the liver histopathology showed that the liver slices of the model group had obvious structural disorder, the nucleus was squeezed, and there were obvious fat vacuoles. The treatment groups improved significantly compared with the model group; ② compared with the normal group, the liver function and blood lipid indexes of the model group increased significantly, while the anti-oxidation indexes decreased significantly. Compared with the model group, each treatment groups were significantly improved; ③ compared with the normal group, the protein and mRNA expression levels of PPARα, ACOX1, CPT-1, and LCAD in the model group were significantly reduced, compared with the model group, those indexes in the treatment groups were significantly up-regulated. This study found that ginkgolide B could regulate the expression of long-chain fatty acid metabolism-related proteins PPARα, ACOX1, CPT-1, and LCAD, meanwhile improve the body's antioxidant capacity, thereby reduce blood lipids, further improve liver function and protect the liver.

ObjectiveTo explore the effect of low-fat and low-energy diet on the ultrasound examination of liver， gallbladder and pancreas of participants. MethodsFrom January 2020 to September 2020，a total of 30 participants were enrolled in this study. Using self-controlled research method， firstly， ultrasound scans were performed to them for scanning some organs （such as pancreas， left liver and gallbladder） after fasting more than 8 hours， then the participants were told to eat the low-fat and energy foods （steamed bread and porridge）， and five times of the same ultrasound scans were performed to them at 0.5 h， 1 h， 2 h， 3 h and 4 h after the meal. Every participant was tested in the above 6 groups. The ultrasound images were blindly evaluated， and gallbladder volumes were compared by two sonographers. The single-factor method of repeated measurement was used to evaluate the ultrasound image qualities of different groups and to compare the gallbladder volumes. ResultsThe difference had no statistical significance in imaging quality of pancreas and left liver between each postprandial group and fasting group （P > 0.05）. Among the most concerned gallbladder ultrasound image quality， the difference had no statistical significance between groups （F=0.7， P=0.6）. For the percentage of gallbladder volume， the difference had no statistical significance between groups （F=1.8， P=0.2）. ConclusionCompared with the ultrasound examination under an 8-hour fasting state， the ultrasound images of abdominal organs （such as liver， pancreas and gallbladder） and the gallbladder volumes are not affected by the low-fat and low-energy diets.