Venot A, Lebruchec J F, Golmard J L, Roucayrol J C
J Nucl Med. 1983 Jun;24(6):529-31.
The computer comparison of two images of the same organ requires normalization of the images. We propose an iterative, automated method of normalization that is highly insensitive to structural modifications in the images. The normalization factor is calculated by maximizing the number of sign changes in the scanned subtraction image with a unidimensionnal optimization method. This algorithm is efficient, computationally cheap, and can be implemented on most nuclear medicine data-processing systems.
对同一器官的两幅图像进行计算机比较需要对图像进行归一化处理。我们提出了一种迭代的、自动化的归一化方法，该方法对图像中的结构变化高度不敏感。通过使用一维优化方法最大化扫描减法图像中符号变化的数量来计算归一化因子。该算法高效、计算成本低，并且可以在大多数核医学数据处理系统上实现。
