A new estimation method of maximum intensity, such as I spta in CW and I sppa in PW, in high intensity focused ultrasound (HIFU) field for surgery was developed based on the measurements of acousic power and radical d...A new estimation method of maximum intensity, such as I spta in CW and I sppa in PW, in high intensity focused ultrasound (HIFU) field for surgery was developed based on the measurements of acousic power and radical distribution of the relative acoustic pressure in the focal plane of the field. The acoustic power is measured by the radiation force balance or by the calorimetry. Using a miniature hydrophone to scan along several radial directions with the same angle intervals through the acoustic axis in the focal plane, several radial relative pressure distribution curves can be measured and thus the special cross section area related to the maximum intensity and be calculated. The quotient of the measured acousitc power to the special cross section area is the maximum acoustic intensity in the field. finally as an example, the acoustic field distribution of a CW system was measured and the maximum acoustic intensity estimated was given.展开更多
Tri-training利用无标签数据进行分类可有效提高分类器的泛化能力,但其易将无标签数据误标,从而形成训练噪声。提出一种基于密度峰值聚类的Tri-training(Tri-training with density peaks clustering,DPC-TT)算法。密度峰值聚类通过类...Tri-training利用无标签数据进行分类可有效提高分类器的泛化能力,但其易将无标签数据误标,从而形成训练噪声。提出一种基于密度峰值聚类的Tri-training(Tri-training with density peaks clustering,DPC-TT)算法。密度峰值聚类通过类簇中心和局部密度可选出数据空间结构表现较好的样本。DPC-TT算法采用密度峰值聚类算法获取训练数据的类簇中心和样本的局部密度,对类簇中心的截断距离范围内的样本认定为空间结构表现较好,标记为核心数据,使用核心数据更新分类器,可降低迭代过程中的训练噪声,进而提高分类器的性能。实验结果表明:相比于标准Tritraining算法及其改进算法,DPC-TT算法具有更好的分类性能。展开更多
基金National Natural Science Foundation ofChina (No.3 9970 2 0 9)
文摘A new estimation method of maximum intensity, such as I spta in CW and I sppa in PW, in high intensity focused ultrasound (HIFU) field for surgery was developed based on the measurements of acousic power and radical distribution of the relative acoustic pressure in the focal plane of the field. The acoustic power is measured by the radiation force balance or by the calorimetry. Using a miniature hydrophone to scan along several radial directions with the same angle intervals through the acoustic axis in the focal plane, several radial relative pressure distribution curves can be measured and thus the special cross section area related to the maximum intensity and be calculated. The quotient of the measured acousitc power to the special cross section area is the maximum acoustic intensity in the field. finally as an example, the acoustic field distribution of a CW system was measured and the maximum acoustic intensity estimated was given.
文摘Tri-training利用无标签数据进行分类可有效提高分类器的泛化能力,但其易将无标签数据误标,从而形成训练噪声。提出一种基于密度峰值聚类的Tri-training(Tri-training with density peaks clustering,DPC-TT)算法。密度峰值聚类通过类簇中心和局部密度可选出数据空间结构表现较好的样本。DPC-TT算法采用密度峰值聚类算法获取训练数据的类簇中心和样本的局部密度,对类簇中心的截断距离范围内的样本认定为空间结构表现较好,标记为核心数据,使用核心数据更新分类器,可降低迭代过程中的训练噪声,进而提高分类器的性能。实验结果表明:相比于标准Tritraining算法及其改进算法,DPC-TT算法具有更好的分类性能。
