FOR three decades, Zhong Jianhua worked in diplomatic relations be- tween China, the United Kingdom and the United States. Then he found himself posted to South Af- rica. This shift away from Western affairs, his area...FOR three decades, Zhong Jianhua worked in diplomatic relations be- tween China, the United Kingdom and the United States. Then he found himself posted to South Af- rica. This shift away from Western affairs, his area of expertise, made his career path a unique one among the Chinese ambassadors normally sent to Africa.展开更多
文摘FOR three decades, Zhong Jianhua worked in diplomatic relations be- tween China, the United Kingdom and the United States. Then he found himself posted to South Af- rica. This shift away from Western affairs, his area of expertise, made his career path a unique one among the Chinese ambassadors normally sent to Africa.
文摘P53基因状态是胶质瘤精准诊疗的重要依据.针对目前基于MRI(Magnetic Resonance Imaging)的P53基因状态预测的深度学习模型中存在的异质性特征提取不全面、模型存在固有的多种不确定性等问题,提出脑胶质瘤P53基因状态精准预测模型CVT-RegNet(Improved RegNet Integrating CNN,Vision Transfomer and Truth Discovery).首先,采用RegNet网络作为P53基因突变状态预测模型的基础架构,自适应设计搜索P53基因的异质性特征;其次,在模型中将ViT(Vision Transfomer)模块与卷积神经网络(Convolutional Neural Networks,CNN)模块进行融合以改进RegNet网络,进一步优化模型的特征提取性能与计算效率;最后,融入真值发现算法进行迭代寻优以改善模型输出的不确定性,提高预测结果的准确度.实验结果表明,CVT-RegNet模型对P53突变状态的预测准确率达到95.06%,AUC(Area under Curve)得分为0.9492,优于现有的P53基因状态预测模型.CVT-RegNet实现了胶质瘤P53基因状态的无创预测,减轻了患者的经济负担及身心伤害,为胶质瘤的临床精准诊断治疗提供了重要价值.
