少样本学习是目前机器学习研究领域的热点和难点.针对现有的少样本学习模型不能有效捕捉数据特征与数据标签之间的联系,造成分类模型泛化能力弱的问题,提出一种基于元学习的原型空间图卷积网络少样本学习模型FSL-GCNPS(Few-Shot Learnin...少样本学习是目前机器学习研究领域的热点和难点.针对现有的少样本学习模型不能有效捕捉数据特征与数据标签之间的联系,造成分类模型泛化能力弱的问题,提出一种基于元学习的原型空间图卷积网络少样本学习模型FSL-GCNPS(Few-Shot Learning of Graph Convolutional Network on Prototype Space).首先,利用卷积神经网络提取多任务数据的特征向量;其次,为了将特征向量映射到原型空间中,根据元学习的训练策略得到特征向量的类原型表达;然后,通过类原型向量和类向量之间的嵌入表示,构建图结构数据,并进行图卷积网络训练、推理.实验结果表明,相较于经典少样本学习方法,FSL-GCNPS模型拥有更好的分类准确率和分类稳定性.同时,在医学图像领域数据集上实验表明,FSL-GCNPS具有很好的跨域适应性.展开更多
BACKGROUND Bile leakage is a common and serious complication of open hepatectomy for the treatment of biliary tract cancer.AIM To evaluate the incidence,risk factors,and management of bile leakage after open hepatecto...BACKGROUND Bile leakage is a common and serious complication of open hepatectomy for the treatment of biliary tract cancer.AIM To evaluate the incidence,risk factors,and management of bile leakage after open hepatectomy in patients with biliary tract cancer.METHODS We retrospectively analyzed 120 patients who underwent open hepatectomy for biliary tract cancer from February 2018 to February 2023.Bile leak was defined as bile drainage from the surgical site or drain or the presence of a biloma on imaging.The incidence,severity,timing,location,and treatment of the bile leaks were recorded.The risk factors for bile leakage were analyzed using univariate and multivariate logistic regression analyses.RESULTS The incidence of bile leak was 16.7%(20/120),and most cases were grade A(75%,15/20)according to the International Study Group of Liver Surgery classification.The median time of onset was 5 d(range,1-14 d),and the median duration was 7 d(range,2-28 d).The most common location of bile leakage was the cut surface of the liver(70%,14/20),followed by the anastomosis site(25%,5/20)and the cystic duct stump(5%,1/20).Most bile leaks were treated conservatively with drainage,antibiotics,and nutritional support(85%,17/20),whereas some required endoscopic retrograde cholangiopancreatography with stenting(10%,2/20)or percutaneous transhepatic cholangiography with drainage(5%,1/20).Risk factors for bile leakage include male sex,hepatocellular carcinoma,major hepatectomy,blood loss,and blood transfusion.CONCLUSION Bile leakage is a frequent complication of open hepatectomy for biliary tract cancer.However,most cases are mild and can be conservatively managed.Male sex,hepatocellular carcinoma,major hepatectomy,blood loss,and blood transfusion were associated with an increased risk of bile leak.展开更多
Laser scanning confocal endomicroscope(LSCEM)has emerged as an imaging modality which provides noninvasive,in vivo imaging of biological tissue on a microscopic scale.Scientific visualizations for LSCEM datasets captu...Laser scanning confocal endomicroscope(LSCEM)has emerged as an imaging modality which provides noninvasive,in vivo imaging of biological tissue on a microscopic scale.Scientific visualizations for LSCEM datasets captured by current imaging systems require these datasets to be fully acquired and brought to a separate rendering machine.To extend the features and capabilities of this modality,we propose a system which is capable of performing realtime visualization of LSCEM datasets.Using field-programmable gate arrays,our system performs three tasks in parallel:(1)automated control of dataset acquisition;(2)imaging-rendering system synchronization;and(3)realtime volume rendering of dynamic datasets.Through fusion of LSCEM imaging and volume rendering processes,acquired datasets can be visualized in realtime to provide an immediate perception of the image quality and biological conditions of the subject,further assisting in realtime cancer diagnosis.Subsequently,the imaging procedure can be improved for more accurate diagnosis and reduce the need for repeating the process due to unsatisfactory datasets.展开更多
文摘少样本学习是目前机器学习研究领域的热点和难点.针对现有的少样本学习模型不能有效捕捉数据特征与数据标签之间的联系,造成分类模型泛化能力弱的问题,提出一种基于元学习的原型空间图卷积网络少样本学习模型FSL-GCNPS(Few-Shot Learning of Graph Convolutional Network on Prototype Space).首先,利用卷积神经网络提取多任务数据的特征向量;其次,为了将特征向量映射到原型空间中,根据元学习的训练策略得到特征向量的类原型表达;然后,通过类原型向量和类向量之间的嵌入表示,构建图结构数据,并进行图卷积网络训练、推理.实验结果表明,相较于经典少样本学习方法,FSL-GCNPS模型拥有更好的分类准确率和分类稳定性.同时,在医学图像领域数据集上实验表明,FSL-GCNPS具有很好的跨域适应性.
文摘BACKGROUND Bile leakage is a common and serious complication of open hepatectomy for the treatment of biliary tract cancer.AIM To evaluate the incidence,risk factors,and management of bile leakage after open hepatectomy in patients with biliary tract cancer.METHODS We retrospectively analyzed 120 patients who underwent open hepatectomy for biliary tract cancer from February 2018 to February 2023.Bile leak was defined as bile drainage from the surgical site or drain or the presence of a biloma on imaging.The incidence,severity,timing,location,and treatment of the bile leaks were recorded.The risk factors for bile leakage were analyzed using univariate and multivariate logistic regression analyses.RESULTS The incidence of bile leak was 16.7%(20/120),and most cases were grade A(75%,15/20)according to the International Study Group of Liver Surgery classification.The median time of onset was 5 d(range,1-14 d),and the median duration was 7 d(range,2-28 d).The most common location of bile leakage was the cut surface of the liver(70%,14/20),followed by the anastomosis site(25%,5/20)and the cystic duct stump(5%,1/20).Most bile leaks were treated conservatively with drainage,antibiotics,and nutritional support(85%,17/20),whereas some required endoscopic retrograde cholangiopancreatography with stenting(10%,2/20)or percutaneous transhepatic cholangiography with drainage(5%,1/20).Risk factors for bile leakage include male sex,hepatocellular carcinoma,major hepatectomy,blood loss,and blood transfusion.CONCLUSION Bile leakage is a frequent complication of open hepatectomy for biliary tract cancer.However,most cases are mild and can be conservatively managed.Male sex,hepatocellular carcinoma,major hepatectomy,blood loss,and blood transfusion were associated with an increased risk of bile leak.
文摘Laser scanning confocal endomicroscope(LSCEM)has emerged as an imaging modality which provides noninvasive,in vivo imaging of biological tissue on a microscopic scale.Scientific visualizations for LSCEM datasets captured by current imaging systems require these datasets to be fully acquired and brought to a separate rendering machine.To extend the features and capabilities of this modality,we propose a system which is capable of performing realtime visualization of LSCEM datasets.Using field-programmable gate arrays,our system performs three tasks in parallel:(1)automated control of dataset acquisition;(2)imaging-rendering system synchronization;and(3)realtime volume rendering of dynamic datasets.Through fusion of LSCEM imaging and volume rendering processes,acquired datasets can be visualized in realtime to provide an immediate perception of the image quality and biological conditions of the subject,further assisting in realtime cancer diagnosis.Subsequently,the imaging procedure can be improved for more accurate diagnosis and reduce the need for repeating the process due to unsatisfactory datasets.