Direction des Relations Internationales (DRI)

Programme INRIA "Equipes Associées"

Please have a look at our three year review.

 

Please have a look at the publication section to see our latest results (updated on 01/2012)

1. Description of the results :

2. List of the exchanges carried out :

3. List of conference and journal papers produced within the context of the Associate Team :

4. SCIENTIFIC WORK PROGRAM FOR NEXT YEAR

Next we will follow up ongoing efforts on 3 different topics:

5. EXCHANGES PROGRAM FOR NEXT YEAR

1. Exchange of researchers

Erin Stretton is planned to make a short visit in Heidelberg (DKFZ) to better understand the tumor database.

 From INRIA to the partner institution
Number of people
Estimated cost
PhD student  1: Erin Stretton  2500 € (2 weeks)
   Total
  1   2500 €

Bjoern Menze will keep spending 25% of his time at the asclepios lab.

 From the partner institution to INRIA
Number of people
Estimated cost
Young researcher (Post-doc)
 1: Bjoern Menze 7500 €  (3 months)
   Total
  1   7500 €

2. Financial contribution from the international partner to the exchange program :

The salary of Bjoern Menze keeps beeing supported by the partner institution during his stay at INRIA.

Please indicate the amount of funding expected from the partner and from external resources :

8500 €

Associate Team budget proposal for year 1 (automated) :

Associate Team budget proposal for year 1 (automated):

Global cost of the collaboration project: 18500 €

External resources (other than Associate Team program): 8500 €

Funding from the Associate Team program: 10000 €


DEFINITION



ASSOCIATE TEAM
CompuTumor
selected in year
2010
Projet INRIA : ASCLEPIOS Organisme étranger partenaire : MIT
Unité de recherche INRIA : INRIA SOPHIA ANTIPOLIS
Thème INRIA : BIO
Pays : ETATS UNIS
 

 
Coordinateurs français
Coordinateur étranger
Nom, prénom Nicholas Ayache, PhD Olivier Clatz, PhD Polina Golland, PhD Bjoern Menze, PhD Bram Stieltjes, MD
Grade/statut Directeur de Recherche Chargé de Recherche Associate Professor Research Associate Associate Professor
Organisme d'appartenance
INRIA Sophia Antipolis, Projet Asclepios Massachusetts Institute of Technology , Computer Science and Artificial Intelligence Laboratory (CSAIL) DKFZ Heidelberg, diffusion group.
Adresse postale INRIA Sophia Antipolis
Asclepios Research Project
2004 route des Lucioles - BP 93
06902 Sophia Antipolis Cedex
France
MIT
CSAIL
32 Vassar Street 32-D429
Cambridge, MA 02139 USA
Department of Radiology, German Cancer Research Center, D-69120 Heidelberg, Germany. m.a.weber@dkfz.de
URL AYACHE CLATZ Golland Menze
Téléphone +33 4 92 38 76 61 +33 4 92 38 71 59 +1-617-253-5860 +1-617-452-2124 +49-6221-422492
Courriel Nicholas.Ayache[at]sophia.inria.fr olivier.clatz[at]sophia.inria.fr fern[at]csail.mit.edu menze[at]csail.mit.edu B.Stieltjes[at]dkfz-heidelberg.de



La proposition en bref

Title of the collaboration theme : Computational Brain Tumor

Summary : The CompuTumor project is dedicated to the study of brain tumor models and their coupling with medical images to better assist diagnosis and therapy. The project will strongly enhance the current collaborations between INRIA and a group of world leading teams with complementary technical and clinical expertise on these topics in Boston and Nice. More specifically, the project aims at (a) proposing new medical image processing method that could be used to better analize tumor images, (b) developping new brain tumor models in order to personalise these models with patient data.

Présentation détaillée de l'Équipe Associée

1. Objectifs scientifiques de la proposition 

The CompuTumor project is dedicated to the study and development of brain tumor models for improved therapy. This project aims at expanding the work of Asclepios in computational brain tumor through the collaboration with world class leaders showing expertise in complementary research domains. In the continuity of the research program proposed in the previous proposal, this new proposal will extend our research along two main themes: (1) Medical image processing method for tumor images (2) Brain tumor growth modeling and personalisation of models. This scope of the project has been reduced to cope with the more limited budget available in the renewal of a team. It should be noted that an additional funding demand involving more partners and having  Thomas Deisboeck as a PI (Harvard) is currently under review by NIH.

1. Medical image processing method for tumor images

Tumor images represent a chalenge for classical medical image processing algorithms. Indeed, state of the art segmentation and registration algorithms fail in the presence of lesion in the images. This weakness is due to the diversity in appearance of tumor in medical images. In particular, glial tumors tend to infiltrate the brain tissue which make the distinct identification of tumor tissue difficult. In this section, we will explore new areas of research to adapt existing tools to tumor images. In particular, the extension of the demons algorithms, as well as the developpment of new segmentation algorithms based on the random forest

2. Brain Tumor Growth Modeling.

The objective of this theme is to improve the existing models and to develop new ones to include 1) the uncertainty on the growth of the tumor, 2) the effect of various therapies and 3) a multi-scale modeling approach from microscopic to macroscopic scales.  In a first stage, the research will be dedicated to the introduction of the notion of uncertainty into the existing tools, and of its representation and visualization in the medical images of a patient. The white matter parameters, so far considered as spatially constant, will be evaluated on specific fiber bundles. This will require identifying fiber clusters within the white matter. We will develop methods to provide confidence intervals into the estimated parameters. This confidence will highly depend on the estimated noise on the data, modeling error as well as error in the segmentation process.  In a second stage we will explore how to model various therapies and how to identify their parameters from time series of medical images: existing microscopic models for radiotherapy and chemotherapy will be reviewed and adapted to the current models. Finally, a link between microscopic and macroscopic models will be explored 

2. Présentation des partenaires 

Polina Golland PhD, is an associate professor in the EECS Department and the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT. she is also an associate member of the Broad Institute. She holds the Distinguished Alumnus (1964) Career Development Chair. Her primary research interest is in developing novel techniques for image analysis and understanding. She works on algorithms that either explore the geometry of the world and the imaging process in a new way or improve image-based inference through statistical modeling of the image data. She is interested in shape modeling and representation, predictive modeling and visualization of statistical models. Her current research focuses on developing statistical analysis methods for characterization of biological processes using images (from MRI to microscopy) as a source of information.

Bjoern Menze PhD, is a post-doctoral researcher in the Medical Vision Group of CSAIL. He is interested in the analysis of spatio-temporal physical processes using probabilistic and functional models which rely on vector-valued image data as the primary source of information. He also works on the design and validation of automated diagnostic systems implementing these models (primarily) in biomedical applications. Bjoern hab been one of the key elements during the first three years of the CompuTumor Team and will remain the pivot of the forecoming collaboration.
Bram Stieltjes MD, is the group leader of the diffusion lab. His research focuses on the development of brain tumor models and the introduction of new MR modalities for tumor detection, diagnostic, and therapy. Bjoern also has an appointment into the radiology department and will provide MR data as well as expertise for the interpretation of the cases.

Fritzsche Klaus is starting in PhD in the valuation of new MR diffusion protocol for brain tumor invasion quantification.

3. Impact 

The three groups bring complementary strengths into the collaboration. While MIT medical vision groups brings image analysis methods, the asclepios groups provides expertise in the modeling of brain tumor and the personalisation of models. The DKFZ group comes with the necessary medical expertise and data needed to reach a high clinical impact.

We expect that the already developped algorithm between MIT and Asclepios for image processing, tumor modeling and paramenter estimation will be extended and evaluated on the large database already available at the DKFZ.

 

 

© INRIA - mise à jour le 12/10/2010