CE45 - Mathématique, informatique, automatique, traitement du signal pour répondre aux défis de la biologie et de la santé

CROss-modal registration in COrrelative Microscopies for VALvulopathy physiological characterisation – CROCOVAL

CROCOVAL CROss-modal registration in COrrelative Microscopies for VALvulopathy physiopathological characterisation

Correlative microscopy makes it possible to combine different scales of observations and different types of content, functional and morphological. To decipher the fundamental mechanisms of dysfunction of the heart valve, these approaches are particularly interesting because of the heterogeneity of cell remodeling involved in the development of the disease.<br />The objective of this project is to develop adapted methods of image data fusion to study the biological mechanisms involved.

Combining different observations on the same sample to gain a better understanding of heart valve dysfunction

There is a wide range of microscopy techniques, each with its own specificities in terms of the type of information collected from a sample and limitations in terms of observation comprehensiveness. Combining different scales of observation and various types of microscopy content, both functional and morphological, by conducting observations with different microscopes on the same sample enables its holistic characterization but poses major technological challenges: How do we relocate the same area of interest when moving the sample from one microscope to another? How do we merge information without loss of quality? How can we ensure accuracy in sample localization when image contents are very different? These challenges are particularly relevant for understanding heart valve dysfunction, one of the leading causes of heart problems. In this context, our project aims to develop a computer vision approach that takes into account confidence in the registration of microscopy data to enable efficient acquisition and analysis of different microscopy types. This will pave the way for a better understanding of this common pathology and the application of these advanced methods in other areas of biomedical research.

As part of this project, we have devised registration algorithms based on points and methods derived from graph theory. Furthermore, we have introduced a robust error estimation method based on multivariate regression. This has allowed us to accurately assess the quality of registration, whether it involves rigid or affine transformations.

In parallel, in collaboration with a company, we have created an innovative device called the Kratoscope. This system captures the auto-fluorescence of samples before each section, thus enabling the construction of a stack of aligned 3D images, bridging 3D imaging and histological sections at the cellular level.

We have also enhanced our software, ec-clemv2, to handle larger images and offer advanced functionalities. The software now allows the creation of custom registration workflows by chaining transformations. It also provides the capability to overlay images in a common space without re-sampling and to apply calculated transformations to specific areas of interest.

In summary, our work has focused on the development of tools and techniques for the integration and understanding of images from various sources in biological sciences. These advancements open up exciting new prospects for research in this field, enabling a more in-depth and precise analysis of biomedical samples.

In addition to promising initial experiments on studying valve deformations under pathological conditions, the developed techniques have found broader applications. Collaborations with the United Kingdom, Israel, and other colleagues from Nantes have allowed us to investigate mechanisms of virus infection in cells and the neurons in our intestines, among other areas of research.

Furthermore, the methods and software developed are now being routinely utilized at the Diamond synchrotron in the United Kingdom, where users are trained in their usage. They are also made available on the Nantes microscopy platform to benefit a wide range of projects. Additionally, a new international scientific society has been established to promote these correlative approaches.

This progress represents a significant step forward in the field, facilitating collaborative research and expanding the scope of applications for these innovative techniques and tools.

Le projet a conduit à des développements méthodologiques novateurs, poussés par la problématique de l’étude des valvulopathies en exemple d’application. Ces développements sont mis à la disposition de la communauté et ont déjà été utilisées pour d’autres problématiques, répondant ainsi à un besoin de logiciel et d’algorithme en soutien aux acquisitions multimodales.

The work carried out in this context has been valorized through research papers (11) and presentations at conferences (16), including ten invited presentations that demonstrate the interest generated by our developments. Additionally, we have made software freely accessible to the public. This project has also enabled our participation in an international standardization project related to certain aspects of these approaches. The complete list of publications associated with this project is available at anrcrocoval.github.io/publications.html.

Correlative microscopies allow to combine different scales of observations and different contents, functional and morphological, based on the large panel of microscopy technologies available for life sciences. Deciphering the fundamental process of cardiac valve dysfunction in particular would greatly benefit from these approaches, because of the heterogeneity of the cellular remodeling processes involved in the development of the disease. In this context, the purpose of this proposal is to reveal this so far unexploited potential of correlative microscopies to unravel fundamental processes causing mitral valve dysfunction by developing adapted data fusion methods. We want to develop an original computer vision approach taking into account the confidence of correlation to acquire and analyse image data from microscopies on an in-house developped animal model of valvulopathy.
We expect that the proposed framework will apply to a larger panel of biomedical studies.

Project coordination

Perrine Paul-Gilloteaux (Unité de recherche de l'Institut du Thorax)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.

Partner

INSERM UMR 1087 / CNRS UMR 6291 Unité de recherche de l'Institut du Thorax

Help of the ANR 223,884 euros
Beginning and duration of the scientific project: March 2019 - 42 Months

Useful links

Explorez notre base de projets financés

 

 

ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter