Registration and Atlas Formation
RESEARCH

The statistical analysis of disease often requires mapping multiple images into a common coordinate system. Our group has been working on a tissue-based, multi-modal, fully deformable, diffeomophic form of registration that maps multi-modal images of all subjects to a central image.

A complementary approach is provided by a new method of 3D-3D, vessel-based registration that uses distance maps to provide a probabilistic vessel atlas. New work that combines vessel probabilities and spatial information suggests that healthy subjects exhibit differences between sexes.

Additional uses of 3D-3D vessel-based registration are summarized in the surgical application link at left. A 3D-2D extension of the registration approach is outlined in the endovascular link at left.

Supported by EB000219 NIBIB-NIH and HL6908 HLB-NIH
Central image produced by affine registration (left) and fully deformable registration (right). The left image is blurred whereas the right is not. The slices are the same. See Dr. Joshi for additional information.
Creation of a distance map used in vessel atlas formation. See Dr. Aylward for additional detail.
REFERENCES

Aylward SR, Jomier J, Weeks S, Bullitt E (2003) Registration and analysis of vascular images. International Journal of Computer Vision. 55:123-138 <pdf>.

Lorenzen P, Davis B, Gerig G, Bullitt E, Joshi S (2004) Multi-class Posterior Atlas Formation via Unbiased Kullback-Leibler Template Estimation. MICCAI 2004; Lecture Notes in Computer Science 3216:95-102 <pdf>.

Aylward S, Jomier J, Vivert C, Ledigarcher V, Bullitt E (2005) Spatial graphs for intracranial vascular network characterization, generation, and discrimination. MICCAI 2005; LNCS 3749: 59-66.<pdf>

Lorenzen P, Prastawa M, Davis B, Gerig G, Bullitt E, Joshi S (2006) Multi-modal image set registration. MedIA 10:440-451.<pdf>

Davis B, Fletcher PT, Bullitt E, Joshi S (2007) Population Shape Regression from Random Design Data. ICCV 2007. Marr Prize winner 2007. <pdf>