摘要：A series of novel 2,6-dichloro-3,5-dinitrotoluene derivatives were designed, synthesized in the present study, and their antitumor activities against five cell lines（A431, HepG2, A549, HT-29 and HEK-293） were tested. Most of the compounds exhibited moderate-to-significant cytotoxicity and high selectivity against one or more cell lines in comparison with cisplatin. Studies on their preliminary structure-activity relationships（SARs） indicated that compounds containing phenyl（piperazin-1-yl） methanone groups, especially chlorine atom at 4-position of the phenyl ring, were more effective. Compound 4g was found to be the most potent derivative with IC_（50） values of 1.04, 3.20, 6.93, 4.10 and 20.15 μmol/L against A431, HepG2, A549, HT-29 and HEK-293 cell lines, respectively, which was better than positive control cisplatin, one of the most clinically used chemotherapeutic drugs.

摘要：图像的特征选择需要筛除大量噪声节点,在像素较高的图像内处理效率低下,为减少图像特征选择的处理时间,设计基于适用性骨干粒子群优化算法的图像特征选择方法。提取图像视觉特征参数,包括颜色参数、纹理参数以及形状参数。获取分类面的线性判别函数,建立最优超平面,得到满足约束条件的目标函数以及特征选择的适应度函数,基于适用性骨干粒子群引入粒子位置与速度更新机制,设计特征选择算法,得到一个新的图像特征选择处理方法。获取不同阈值以及不同学习效率下的最优特征数量,分别测试四种数据集内特征选择算法的运行时间,实验结果显示,在四种数据集内,适用性骨干粒子群优化算法的运行时间均小于其他算法,可见该算法为相同图像相同参数下的最优算法。