数理统计在医学图像配准评价中的应用

Application of Statistical Method to Evaluate Medical Image Registration

目的研究利用数理统计方法对图像配准算法进行评估的方法。 方法对于图像在配准前后的位置、形态及灰度值变化分别用 χ2 检验和 μ检验进行统计推断。 结果通过统计推断的结论可以判断所用配准方法是否有效。 结论用数理统计方法能对配准方法进行客观、正确地评价。

Objective To evaluate image registration algorithm by using the statistical method.Methods The χ 2 test and μ test were used to infer statistically the changes of image position, morphology and gray value between the original and registrated images. Results The effect of image registration algorithm was judged by statistical inference. Conclusion The result shows that the statistical method could evaluate registration algorithm objectively and correctly.