Overall Objectives
Research Program
New Software and Platforms
New Results
Bilateral Contracts and Grants with Industry
Partnerships and Cooperations
XML PDF e-pub
PDF e-Pub

Section: New Results

Medical Image Analysis

3D/2D Coronary Arteries Registration

Participants : Thomas Benseghir [correspondent] , Grégoire Malandain [Morpheme Team] , Régis Vaillant [GE-Healthcare] , Nicholas Ayache.

This work has been performed in colaboration with GE-Healthcare (Buc) and the Morpheme team at Inria SAM.

3D/2D Registration, Computed Tomography Angiography, X-ray Fluoroscopy, Coronary Arteries, Vascular Tree

Integrating vessel calcifications and occlusion information, extracted from pre-operative 3D CT angiography images into a live fluoroscopic 2D image can greatly improve the guidance of percutaneous coronary interventions. Such task requires a registration step that must provide relevant correspondences between these two complementary modalities. We are developing a framework aiming at preserving the topology of the vascular structures matched between both images.

The introduction of topology in the pairing procedure allows to decrease the mismatches with respect to geometrically-based pairing procedures (e.g. Iterative Closest Point), which, in turn, improves the success rate of the registration method. This is exemplified by Fig. 3 where the proposed pairing method is compared to ICP.

Figure 3. From left to right: 1. initial pose estimate (3D centerlines projection in blue); 2. iterative closest point algorithm registered position (point-pairings in yellow); 3: proposed method registered position; 4: resulting fusion between the two modalities
IMG/benseghirinitialPose.png IMG/benseghirregisteredPoseICP.png IMG/benseghirregisteredPoseICC.png IMG/benseghirFusion.png

Video Synchronization: An Approach Towards Endoscopic Re-localization

Participants : Anant Suraj Vemuri [correspondent] , Nicholas Ayache.

Endoscopy, Barrett's Esophagus, Re-localization, Electromagnetic Tracking

Figure 4. (Left) The orange circular markings indicate the position of the sensors taken as anatomical landmarks to form the reference frame for the patient. (Right) System setup in the operating room.

A sparse Bayesian framework for non-rigid registration

Participants : Loic Le Folgoc [correspondent] , Hervé Delingette, Antonio Criminisi, Nicholas Ayache.

This work has been partly supported by Microsoft Research - Inria joint laboratory through its PhD Scholarship Programme and the European Research Council through the ERC Advanced Grant MedYMA (on Biophysical Modeling and Analysis of Dynamic Medical Images).

Registration, Automatic Relevance Determination, Uncertainty Quantification

We propose a sparse Bayesian framework for non-rigid registration. It provides a principled approach to efficiently find an optimal, sparse parameterization of deformations among any preset, widely overcomplete range of basis functions. It addresses open challenges in state-of-the-art registration, such as the automatic joint estimate of model parameters (e.g. noise and regularization levels). We have evaluated the feasibility and performance of our approach on cine MR, tagged MR and 3D US cardiac images, and show state-of-the-art results on benchmark datasets evaluating accuracy of motion and strain (see Fig.5 ). This work was presented during the MICCAI 2014 conference[20] .

Figure 5. (Left) Mesh contour propagated to end systole via the registration output; (Right) Spatial uncertainty visualized as a tensor map.
IMG/lefolgocFig1.png IMG/lefolgocFig2.png

Segmentation and anatomic variability of the cochlea and other temporal bone structures from medical images

Participants : Thomas Demarcy [correspondent] , Hervé Delingette, Clair Vandersteen, Dan Gnansia [Oticon Medical] , Nicholas Ayache.

This work is funded by a CIFRE grant involving Oticon Medical (Vallauris) and performed in collaboration with the IUFC Nice (Pr. Guevara) and CHU Nice (Pr. Raffaelli) .

image segmentation ; surgery planning ; shape modelling ; anatomic variability ; cochlear implant ; temporal bone

Figure 6. Virtual view of the posterior tympanotomy approach fused with a CT-scan

Understanding cardiac planes of acquisition

Participants : Jan Margeta [correspondent] , Nicholas Ayache, Daniel C Lee [Northwestern University] , Antonio Criminisi [Microsoft Research Cambridge] .

This work has been partly supported by Microsoft Research through its PhD Scholarship Programme, by ERC Advanced Grant MedYMA (on Biophysical Modeling and Analysis of Dynamic Medical Images), and by the VP2HF FP7 research project.

Cardiac imaging, Machine learning, Magnetic resonance, Data wrangling

DICOM image format defines several tags by which the images can be queried and filtered. Many useful tags are however not standardized and must be cleaned prior to any large scale analysis.

Figure 7. Examples of cardiac acquisition plane predictions and confidences