A Fast Algorithm to Estimate Inverse Consistent Image Transformation Based on Corresponding Landmarks

Image transformation is widely used to map images, and plays an important role in imaging analysis such as image registration, animation and shape atlases construction. In image registration, one of the key issues is the transformation model used between the source image and the target image. Transformations between two images are expected to be one-to-one to make sure that each point in the source image just has one corresponding point in the target image and vice versa. In particular, for medical image registration, symmetry and inverse consistency are two commonly required properties of transformations [1] to produce the kind of deformation results, which are smoother and present more biological meaning.

Source: Computerized Medical Imaging and Graphics - August 8, 2015 Category: Radiology Authors: Xuan Yang, Dian Zhang, Shuiyong Yao, Bo Wang Source Type: research