深度学习算法在腭中缝成熟度分期中的应用

目的:探讨深度卷积神经网络模型(CNN)在评估锥形束计算机断层扫描(CBCT)影像中的腭中缝成熟度的应用,验证深度学习算法的有效性。方法:在已有卷积神经网络Xception模型的基础上,对模型进行针对性的结构优化,引入了注意力以及多特征融合机制。使用661例CBCT的腭平面截图,图像经预处理后作为训练集,在对网络模型进行训练后,利用20例典型分期样本进行验证,再分别测试38例困难样本(测试集A)和60例平均难度样本(测试集B)的分期准确率。最后将模型与医生的判断结果进行对比分析。结果:所设计的深度神经网络模型在数据集A和数据集B上的准确率分别为0.868和0.916,医生在数据集A和数据集B的准确率分别为0.628和0.850,结论:在病例出现多期性时,深度神经网络模型能够给出更加准确的结论,因此深度神经网络模型能够给医生提供有价值的参考,辅助医生做出正确的诊断。     

基于加权基因共表达网络的膀胱尿路上皮癌预后及疗效相关基因筛选

目的构建加权基因共表达网络分析（weighed gene co-expression network analysis,WGCNA）,鉴定出膀胱尿路上皮癌（bladder urothelial carcinoma,BLCA）中与预后及治疗效果相关的基因,并分析其在肿瘤相关通路中的活性。方法在癌症基因组图谱数据库获取膀胱尿路上皮癌的转录组测序及相应临床数据;选择年龄、性别、身高、体重、人种、生存时间、吸烟年数、治疗效果、病理分级、TNM分期及淋巴结浸润为临床表型进行加权基因共表达网络分析;利用R语言进行生存分析;利用GSCALite工具进行肿瘤相关的通路活性分析;利用基因表达汇编数据库中的GSE13507和GSE3167数据集进行验证。结果本研究鉴定出与预后及治疗效果均显著相关的基因模块及模块关键基因GPHN、CPSF2、PRMT5、EIF2S1、KLC1、SLC39A9、SRP54及CNIH,上述8个基因在肿瘤相关通路中有一定活性。结论GPHN、CPSF2、PRMT5、EIF2S1、KLC1、SLC39A9、SRP54及CNIH基因在BLCA中与预后及治疗效果均显著相关,并且参与了肿瘤相关通路,可能成为BLCA的生物标志物或潜在治疗靶点。     

Expression of epithelial and neural antigens in small cell and non small cell lung carcinoma

Seventy-one lung carcinomas from 66 different patients were stained with a panel of monoclonal antibodies. Twenty-nine were small cell lung carcinoma (SCLC), 15 adenocarcinomas, 17 squamous carcinomas and 10 large cell carcinomas. Three of the monoclonal antibodies recognize different cytokeratins, three recognize other epithelial antigens and one recognizes a neural antigen. Both formalin-fixed and cryopreserved tumours were studied using an indirect immunoperoxidase method. 23/29 SCLC reacted with all but one of the antibodies which recognize epithelial antigens. This staining was similar to that seen in non small cell lung carcinomas (NSCLC) and provides further evidence that SCLC are true epithelial tumours. All but one of the SCLC stained with the antibody recognizing a neural antigen. This antibody did not stain squamous or adenocarcinomas. However, four of the large cell carcinomas stained well with this antibody, suggesting that SCLC and some large cell carcinomas share a common pathway of differentiation. There were variations of staining seen both within and between tumours. This has obvious implications if immunotargetting with monoclonal antibodies is to be used diagnostically or therapeutically.     

国人延髓动脉的观察

本文观察了110例成人脑标本的延髓外部动脉;采用动脉X线造影法和透明法,在20例人脑标本上观察了延髓内部动脉的形态和供应。将延髓表面分为四区,供应各区的外部动脉可分为相应的4群。前正中群和前外侧群动脉主要来自脊髓前动脉,前正中群动脉供应舌下神经核、内侧丘系及部分锥体束等中缝两侧结构;前外侧群动脉供应大部分锥体束;外侧群动脉来自椎动脉、小脑下后动脉、基底动脉和小脑下前动脉,主要供应网状结构、脊髓丘脑束、三叉神经脊束及核、迷走神经背核等;后群动脉来自脊髓后动脉和小脑下后动脉,主要供应薄、楔束核等。本文还讨论了延髓外侧区的血液供应及“终动脉”等问题。     

Automated Computational Detection of Interstitial Fibrosis,Tubular Atrophy,and Glomerulosclerosis

Open in a separate window     

A single‐arm trial indirect comparison investigation: a proof‐of‐concept method to predict venous leg ulcer healing time for a new acellular synthetic matrix matched to standard care control

To compare data on time to healing from two separate cohorts: one treated with a new acellular synthetic matrix plus standard care (SC) and one matched from four large UK pragmatic, randomised controlled trials [venous leg ulcer (VLU) evidence network]. We introduce a new proof‐of‐concept strategy to a VLU clinical evidence network, propensity score matching and sensitivity analysis to predict the feasibility of the new acellular synthetic matrix plus SC for success in future randomised, controlled clinical trials. Prospective data on chronic VLUs from a safety and effectiveness study on an acellular synthetic matrix conducted in one wound centre in the UK (17 patients) and three wound centres in Australia (36 patients) were compared retrospectively to propensity score‐matched data from patients with comparable leg ulcer disease aetiology, age, baseline ulcer area, ulcer duration, multi‐layer compression bandaging and majority of care completed in specialist wound centres (average of 1 visit per week), with the outcome measures at comparable follow‐up periods from patients enrolled in four prospective, multicentre, pragmatic, randomised studies of venous ulcers in the UK (the comparison group; VLU evidence network). Analysis using Kaplan–Meier survival curves showed a mean healing time of 73·1 days for ASM plus SC (ASM) treated ulcers in comparison with 83·5 days for comparison group ulcers treated with SC alone (Log rank test, χ² 5·779, P = 0·016) within 12 weeks. Sensitivity analysis indicates that an unobserved covariate would have to change the odds of healing for SC by a factor of 1·1 to impact the baseline results. Results from this study predict a significant effect on healing time when using a new ASM as an adjunct to SC in the treatment of non‐healing venous ulcers in the UK, but results are sensitive to unobserved covariates that may be important in healing time comparison.     

Online identifier–actor–critic algorithm for optimal control of nonlinear systems

In this paper, a novel identifier–actor–critic optimal control scheme is developed for discrete‐time affine nonlinear systems with uncertainties. In contrast to traditional adaptive dynamic programming methodology, which requires at least partial knowledge of the system dynamics, a neural‐network identifier is employed to learn the unknown control coefficient matrix working together with actor–critic‐based scheme to solve the optimal control online. The critic network learns the approximate value function at each step. The actor network attempts to improve the current policy based on the approximate value function. The weights of all neural networks are updated at each sampling instant. Lyapunov theory is utilized to prove the stability of closed‐loop system. It shows that system states and neural network weights are uniformly ultimately bounded. Finally, simulations are provided to illustrate the effectiveness of the developed method. Copyright © 2016 John Wiley & Sons, Ltd.     

On‐farm Study of Human Contact Networks to Document Potential Pathways for Avian Influenza Transmission between Commercial Poultry Farms in Ontario,Canada

Human movements associated with poultry farming create contact networks that might facilitate transmission of avian influenza (AI) between farms during outbreaks. In Canada, no information is available about how these networks connect poultry farms. The purpose of this study was to document human contacts between commercial poultry farms in Ontario, Canada, to learn how AI might be transmitted during outbreaks. We used face‐to‐face interviews with people entering the farm biosecurity perimeter on four layer, one turkey and three broiler breeder poultry farms in Ontario to collect information on between‐farm contacts and biosecurity practices. Over a four‐day study period on each farm, a median of 10.5 people entered the farm biosecurity perimeter (range 2–31). Ninety‐six per cent (111/118) of people consented to be interviewed. Of these, fifty‐three per cent (59/111) had contact with one or more (median 2, degree range 1–14) other poultry farms within 72 h. A median of 25 (range 7–65) human contacts linked study farms to other poultry farms. The mean distance of between‐farm contacts was 53 km. Eighty‐six per cent of people who answered the biosecurity questions (94/109) reported using one or more biosecurity practices. However, on 7/8 farms, at least one person reported that they did not use any biosecurity practices. Fifty per cent of social visitors used biosecurity, whereas 96% of all other people used biosecurity. Ninety‐two per cent of people that entered the poultry barns (46/50) used one or more biosecurity practices, whereas 81% of people (48/59) that did not enter the poultry barns used one or more biosecurity practices. Because our study documented farm visitors who did not use any biosecurity practices and moved between commercial poultry farms, we suggest that rapid trace‐out of human movements is as important as containment zoning to limiting disease spread during an outbreak of highly pathogenic AI in Ontario.