Delgorge Cécile, Rosenberger Christophe, Poisson Gérard, Vieyres Pierre
Laboratoire Vision and Robotique, University of Orléans, Bourges 18020, France.
IEEE Trans Med Imaging. 2006 Nov;25(11):1502-9. doi: 10.1109/TMI.2006.883088.
This paper presents a new tool for the evaluation of ultrasound image compression. The goal is to measure the image quality as easily as with a statistical criterion, and with the same reliability as the one provided by the medical assessment. An initial experiment is proposed to medical experts and represents our reference value for the comparison of evaluation criteria. Twenty-one statistical criteria are selected from the literature. A cumulative absolute similarity measure is defined as a distance between the criterion to evaluate and the reference value. A first fusion method based on a linear combination of criteria is proposed to improve the results obtained by each of them separately. The second proposed approach combines different statistical criteria and uses the medical assessment in a training phase with a support vector machine. Some experimental results are given and show the benefit of fusion.
本文提出了一种用于评估超声图像压缩的新工具。目标是像使用统计标准一样轻松地测量图像质量，并具有与医学评估相同的可靠性。我们向医学专家提出了一项初步实验，该实验代表了我们用于评估标准比较的参考值。从文献中选择了21种统计标准。定义了一种累积绝对相似性度量，作为待评估标准与参考值之间的距离。提出了一种基于标准线性组合的第一种融合方法，以改进分别由每个标准获得的结果。提出的第二种方法结合了不同的统计标准，并在训练阶段使用支持向量机结合医学评估。给出了一些实验结果，显示了融合的益处。
