Service interruption on Monday 11 July from 12:30 to 13:00: all the sites of the CCSD (HAL, Epiciences, SciencesConf, AureHAL) will be inaccessible (network hardware connection).
Skip to Main content Skip to Navigation

Outils d'imagerie multimodalité pour la neuroradiologie interventionnelle

Abstract : Interventional neuroradiologists make use of diverse imaging modalities: digital subtracted angiography (DSA), 3D angiography (3DXA) and magnetic resonance imaging (MRI). The physician will fully beneficiate from their complementarity only when these modalities are registered. This thesis suggests solutions to the DSA/ 3DXA and 3DXA/ MRI registration problems that do not require any segmentation step. The registration of a DSA image with a 3DXA volume is considered as the search for the parameters of a conic projection perturbed by a distortion field. Procedures are described to calibrate the angiography machine that allow to recover the distortion field by interpolation and enable a precise estimation of the intrinsic parameters of the conic projection. The DSA/ 3DXA registration problem is then equivalent to the extrinsic parameters estimation. An original algorithm is described to recover them byb comparing the DSA .image to the maximum intensity projection (MIP) of the 3DXA volume. An initial match is reached by maximizing the correlation. The result is then refined through an iterative procedure based on the optical flow calculation. Lastly, the 3DXA/MRI registration is studied. The main obstacle lies in the huge difference between both signals. In order to overcome this, we devised an approach based on the modular definition of a voxel similarity measure. It consists in filtering the data to make them more similar, the comparison being then performed thanks to a linear correlation criterion. Applying this strategy to the 3DXA/MRI registration leads to an original, fast and precise algorithm. This algorithm has been validated by comparing it to a fully manual registration.
Document type :
Complete list of metadata
Contributor : Thèses UL Connect in order to contact the contributor
Submitted on : Thursday, March 29, 2018 - 12:33:24 PM
Last modification on : Friday, November 12, 2021 - 11:28:06 AM
Long-term archiving on: : Friday, September 14, 2018 - 10:52:12 AM


Files produced by the author(s)


  • HAL Id : tel-01749938, version 1



Erwan Kerrien. Outils d'imagerie multimodalité pour la neuroradiologie interventionnelle. Autre [cs.OH]. Institut National Polytechnique de Lorraine, 2000. Français. ⟨NNT : 2000INPL003N⟩. ⟨tel-01749938⟩



Record views


Files downloads