Reconstruction of a 3-D volume from a series of 2-D images and fusion with a 3-D imaging modality

Reconstruction of a 3-D volume from a series of 2-D images

The problem is as follows: given a series of 2-D sections, we want to reconstruct a 3-D volume.

Putting all the 2-D sections into a single stack usually does not yield a good reconstruction.

The co-registration of each couple of consecutives sections allows to recover a first smooth 3-D geometry.

S. Ourselin, A. Roche, G. Subsol, X. Pennec, and N. Ayache. Reconstructing a 3D Structure from Serial Histological Sections. Image and Vision Computing, 19(1-2):25-31, January 2001.

Due to the acquisition process, intensity inhomogeneities may occur from section to section. To compensate for them, a dedicated approach has been developed.

Grégoire Malandain and Eric Bardinet. Intensity compensation within series of images. In Randy E. Ellis and Terry M. Peters, editors, Medical image computing and computer-assisted intervention (MICCAI 2003), volume 2879 of LNCS, Montreal, Canada, pages 41--49, November 2003. Springer Verlag.

Fusion with a 3-D imaging modality

The 3-D registration of the above 3-D reconstruction with a 3-D image that is assumed to represent the true anatomy does not yield satisfactory results. This is mainly due to the fact that the above reconstruction process is unable to retrieve the curvature or the torsion of a 3-D object.

G. Malandain and E. Bardinet. Fusion of autoradiographies with an MR volume using 2-D and 3-D linear transformations. In Chris Taylor and Alison Noble, editors, Proceedings of Information Processing in Medical Imaging (IPMI'03), volume 2732 of LNCS, pages 487--498, 2003. Springer.

	Independent 2-D linear transformations obtained by co-registering the 2-D sections with their corresponding 2-D resampled MR slice allows to retrieve the missed curvature and/or torsion. However, the geometrical coherency of the 3-D reconstructed volume is lost.
	To address this problem, an adequate filtering of the 2-D transformations allows to recover this geometrical coherency.
	By iterating the 3-D and 2-D co-registrations, one is then able to correctly fuse the 3-D image with the reconstruction.

Contact: Grégoire Malandain