Nonrigid Image Registration in Digital Subtraction Angiography Using Multilevel B‐Spline
M Nejati, S Sadri, R Amirfattahi - BioMed research international, 2013 - Wiley Online Library
M Nejati, S Sadri, R Amirfattahi
BioMed research international, 2013•Wiley Online LibraryWe address the problem of motion artifact reduction in digital subtraction angiography (DSA)
using image registration techniques. Most of registration algorithms proposed for application
in DSA, have been designed for peripheral and cerebral angiography images in which we
mainly deal with global rigid motions. These algorithms did not yield good results when
applied to coronary angiography images because of complex nonrigid motions that exist in
this type of angiography images. Multiresolution and iterative algorithms are proposed to …
using image registration techniques. Most of registration algorithms proposed for application
in DSA, have been designed for peripheral and cerebral angiography images in which we
mainly deal with global rigid motions. These algorithms did not yield good results when
applied to coronary angiography images because of complex nonrigid motions that exist in
this type of angiography images. Multiresolution and iterative algorithms are proposed to …
We address the problem of motion artifact reduction in digital subtraction angiography (DSA) using image registration techniques. Most of registration algorithms proposed for application in DSA, have been designed for peripheral and cerebral angiography images in which we mainly deal with global rigid motions. These algorithms did not yield good results when applied to coronary angiography images because of complex nonrigid motions that exist in this type of angiography images. Multiresolution and iterative algorithms are proposed to cope with this problem, but these algorithms are associated with high computational cost which makes them not acceptable for real‐time clinical applications. In this paper we propose a nonrigid image registration algorithm for coronary angiography images that is significantly faster than multiresolution and iterative blocking methods and outperforms competing algorithms evaluated on the same data sets. This algorithm is based on a sparse set of matched feature point pairs and the elastic registration is performed by means of multilevel B‐spline image warping. Experimental results with several clinical data sets demonstrate the effectiveness of our approach.
![](/scholar/images/qa_favicons/wiley.com.png)
以上显示的是最相近的搜索结果。 查看全部搜索结果