基于霍特林变换的三维医学图像快速配准算法

Fast 3-D medical image registration algorithm using Hotelling transform

提出了一种新的基于霍特林变换的三维医学图像快速配准算法,这是将数据压缩技术用于图像配准的一种创新性尝试。传统的基于灰度的方法需要考虑整个三维数据的灰度信息,计算复杂度大,无法满足临床需要。论文将Otus算法与互信息量技术相结合提出了一种新的图像分割算法,用于提取待配准物体,从而得到物体的向量表示;然后通过霍特林变换的平移和旋转性质完成配准。实验结果表明此方法能准确,快速地处理图像刚性配准问题,特别适用于三维医学图像的配准。

This paper presents a new 3-D image registration method based on the Hotelling transform.Compared with intensity based registration methods using the whole volume intensity information,the approach utilizes Hotelling transform to align images. The authors evaluate the effectiveness of this approach by applying it to the simulated and real brain image data（MR,CT,PET, and SPECT）.The experimental results indicate that the algorithm is effective,especially for 3-D medical images.