Two hundred cates of nasopharyngeal carcinoma (NPC) admitted to this department from Feb. 1985 to May. 1988 were analysed according to the CT scanning and clinical findings of the primary lesions prior to radiotherapy...Two hundred cates of nasopharyngeal carcinoma (NPC) admitted to this department from Feb. 1985 to May. 1988 were analysed according to the CT scanning and clinical findings of the primary lesions prior to radiotherapy. The results showed that involvement of parapharyngeal space was very common in NPC, about 80% (160/200 cases) ; particularly unilateral or bilateral retro-styloid spaces, about 69.5% (139/200 cases). It was proposed that patients with NPC had a high Incidence of ipsilateral cervical node metastasis. Contralateral cervical node metastasis was rare. The development of cervical node metastasto in NPC has two modes: one Is direct Infiltration of the retro-stylold space by the lesion; the other Is along the nasopharyngeal lymphatic rete. The data also showed that patients with NPC who presented symptoms of Ⅸ- Ⅲ cranial nerve paralyses always had ipsilateral or bilateral retro- styloid space Infiltrations.展开更多
现有复杂网络关键节点识别方法中缺少对节点本身特征的研究,存在网络拓扑信息提取不全面、特征冗余、泛化性低等问题.为了解决上述问题,本文提出一种基于图结构学习的复杂网络关键节点识别方法.首先,针对网络拓扑信息提取不全面问题,结...现有复杂网络关键节点识别方法中缺少对节点本身特征的研究,存在网络拓扑信息提取不全面、特征冗余、泛化性低等问题.为了解决上述问题,本文提出一种基于图结构学习的复杂网络关键节点识别方法.首先,针对网络拓扑信息提取不全面问题,结合复杂网络微观结构和宏观结构构造节点特征;其次,针对特征冗余问题,提出一个融合选择性状态空间模型(State Space Models)和自监督学习的节点特征提取方法;最后,针对泛化性低问题,利用图结构学习在模型训练层面优化损失函数提高分类精度.利用4个公开数据集上进行了广泛实验,本文方法优于次优方法4.66%,节点分辨率保持稳定.实验表明,所提出方法能有效的识别不同网络的关键节点.展开更多
文摘Two hundred cates of nasopharyngeal carcinoma (NPC) admitted to this department from Feb. 1985 to May. 1988 were analysed according to the CT scanning and clinical findings of the primary lesions prior to radiotherapy. The results showed that involvement of parapharyngeal space was very common in NPC, about 80% (160/200 cases) ; particularly unilateral or bilateral retro-styloid spaces, about 69.5% (139/200 cases). It was proposed that patients with NPC had a high Incidence of ipsilateral cervical node metastasis. Contralateral cervical node metastasis was rare. The development of cervical node metastasto in NPC has two modes: one Is direct Infiltration of the retro-stylold space by the lesion; the other Is along the nasopharyngeal lymphatic rete. The data also showed that patients with NPC who presented symptoms of Ⅸ- Ⅲ cranial nerve paralyses always had ipsilateral or bilateral retro- styloid space Infiltrations.
文摘现有复杂网络关键节点识别方法中缺少对节点本身特征的研究,存在网络拓扑信息提取不全面、特征冗余、泛化性低等问题.为了解决上述问题,本文提出一种基于图结构学习的复杂网络关键节点识别方法.首先,针对网络拓扑信息提取不全面问题,结合复杂网络微观结构和宏观结构构造节点特征;其次,针对特征冗余问题,提出一个融合选择性状态空间模型(State Space Models)和自监督学习的节点特征提取方法;最后,针对泛化性低问题,利用图结构学习在模型训练层面优化损失函数提高分类精度.利用4个公开数据集上进行了广泛实验,本文方法优于次优方法4.66%,节点分辨率保持稳定.实验表明,所提出方法能有效的识别不同网络的关键节点.
