Award for Excellence in Technology Transfer in Computer Vision
by Bernard Hidoine
Project EPIDAURE of INRIA Sophia Antipolis was given a "Recognition Award for Excellence in Transfer of Technology in Computer Vision" for its work in the field of medical images analysis. The prize was received by Nicholas Ayache, scientific director of the EPIDAURE project, during the European Conference on Pattern Recognition in Vienna, Austria.
This prize is related to the design and development of three software products which perform the registration of medical images: Crest-Match, Cardio-Match and Voxel-Match.
Crest-Match carries out the registration of monomodal medical images by extracting crest lines on anatomical surfaces. Principal authors: Jean-Philippe Thirion and Alexis Gourdon
Voxel-Match carries out the fusion of volumeric data obtained by various medical imaging modalities (US, MRI, CT...) using physical energy minimisation methods. Principal author: Gregoire Malandain
Cardio-Match carries out the registration of 3D cardiac images obtained by SPECT for myocardial perfusion studies. It works by mesh fitting under a volumic spline transform to register and align the images on a template, making possible intra- and inter-patient comparisons and quantitative analysis. It is planned to be integrated into Focus Medical products. Principal authors: Jérôme Declerck, Jacques Feldmar, Grégoire Malandain and Fabienne Betting.
Crest-Match and Voxel-Match have been integrated into a software product by Focus Medical, a Grenoble company. It was sold or distributed to more than 20 international hospitals and institutes.
Analysis tools of lesions evolution. Here are several pictures of a same patient suffering from an evolutive sclerosis. By registration and substraction of images, the evolution appears. The white ring is a growing lesion while the black ring is a decreasing one. The rigid registration program gave lead to the product CrestMatch, commercialized by FOCUS medical (left). This is to be compared with the same result without registration, with only a manual selection of the best 2D slice corresponding to the original one (right).
Please contact:
Nicholas Ayache - INRIA
Tel: +33 93 65 76 61
E-mail: Nicholas.Ayache@inria.fr
