目的:基于多参数磁共振成像(magnetic resonance imaging,MRI)及多算法机器学习模型识别原发性中枢神经系统淋巴瘤(primary central nervous system lymphoma,PCNSL)中的Bcl-2和c-Myc双表达淋巴瘤(double-expression lymphoma,DEL)。方...目的:基于多参数磁共振成像(magnetic resonance imaging,MRI)及多算法机器学习模型识别原发性中枢神经系统淋巴瘤(primary central nervous system lymphoma,PCNSL)中的Bcl-2和c-Myc双表达淋巴瘤(double-expression lymphoma,DEL)。方法:回顾并分析2021年1月—2024年1月茂名市人民医院及信宜市人民医院颅内PCNSL患者的临床资料,根据免疫组织化学染色法检测Bcl-2和c-Myc蛋白的表达分为DEL组和非共表达组(nDEL组)。对T1加权对比增强成像、弥散加权成像及水抑制成像进行手工分割肿瘤并提取影像组学特征。应用重复最小绝对值收缩和选择算子筛选特征,然后使用15种机器学习算法及参数调整、自定义参数组合、10折交叉验证构建鉴别模型。结果:共纳入200例颅内PCNSL患者(81例DEL)。训练集、内部验证集和外部验证集之间年龄、性别、是否出血坏死、肿瘤位置、瘤周水肿、肿瘤最大径、肿瘤数量、是否侵犯脑膜或室管膜差异无统计学意义。根据组内相关系数(>0.75)初步获取2895个稳定性影像学特征,筛选得到16个特征,极端梯度提升模型及梯度提升机模型鉴别效能最优,验证集最高曲线下面积(area under curve,AUC)为0.91。结论:基于多参数MRI及多算法机器学习模型在DEL检测中有良好的应用前景。展开更多
To measure the void fraction online in oil-gas pipeline, an improved electrical capacitance tomography (ECT) system has been designed. The capacitance sensor with new structure has twelve internal electrodes and overc...To measure the void fraction online in oil-gas pipeline, an improved electrical capacitance tomography (ECT) system has been designed. The capacitance sensor with new structure has twelve internal electrodes and overcomes the influence of the pipe wall. The data collection system is improved by using high performance IC (integrated circuit). Static tests of bubble flow, stratified flow and annular flow regime are carried out. Measurements are taken on bubble flow, stratified flow and slug flow. Results show that the new ECT system performs well on void fraction measurement of bubble flow and stratified flow, but the error of measurement for slug flow is more than 10%.展开更多
文摘目的:基于多参数磁共振成像(magnetic resonance imaging,MRI)及多算法机器学习模型识别原发性中枢神经系统淋巴瘤(primary central nervous system lymphoma,PCNSL)中的Bcl-2和c-Myc双表达淋巴瘤(double-expression lymphoma,DEL)。方法:回顾并分析2021年1月—2024年1月茂名市人民医院及信宜市人民医院颅内PCNSL患者的临床资料,根据免疫组织化学染色法检测Bcl-2和c-Myc蛋白的表达分为DEL组和非共表达组(nDEL组)。对T1加权对比增强成像、弥散加权成像及水抑制成像进行手工分割肿瘤并提取影像组学特征。应用重复最小绝对值收缩和选择算子筛选特征,然后使用15种机器学习算法及参数调整、自定义参数组合、10折交叉验证构建鉴别模型。结果:共纳入200例颅内PCNSL患者(81例DEL)。训练集、内部验证集和外部验证集之间年龄、性别、是否出血坏死、肿瘤位置、瘤周水肿、肿瘤最大径、肿瘤数量、是否侵犯脑膜或室管膜差异无统计学意义。根据组内相关系数(>0.75)初步获取2895个稳定性影像学特征,筛选得到16个特征,极端梯度提升模型及梯度提升机模型鉴别效能最优,验证集最高曲线下面积(area under curve,AUC)为0.91。结论:基于多参数MRI及多算法机器学习模型在DEL检测中有良好的应用前景。
基金the National High Technology Research and Development Program of China (863 Program) (No. 2002AA616050).
文摘To measure the void fraction online in oil-gas pipeline, an improved electrical capacitance tomography (ECT) system has been designed. The capacitance sensor with new structure has twelve internal electrodes and overcomes the influence of the pipe wall. The data collection system is improved by using high performance IC (integrated circuit). Static tests of bubble flow, stratified flow and annular flow regime are carried out. Measurements are taken on bubble flow, stratified flow and slug flow. Results show that the new ECT system performs well on void fraction measurement of bubble flow and stratified flow, but the error of measurement for slug flow is more than 10%.
