Motion estimation and structure tracking in echocardiography using specific images: the UltraSound Tagging. – US-Tagging

The proposed approach consists in rethinking the way the ultrasound images are formed. Whereasmost of the studies are based on the use of conventional images, this projects goes until changing the way the images are constructed by the ultrasound scanner. Then, thanks to an identification and tracking of the cardiac structures, the clinicians will be offered new diagnostic.

The main results achived till now are:
1) the integration of the new image formation technique adapted for motion estimation of the heart. This is the first time that the such images have been shown in real time.
2) the second important result concerns the development of simulation tools for the validation of the technique. Any technique needs images where the parameters are perfectly controlled and known, in order to validate quantitatively their results. This kind of images is often hard to obtain experimentally. On the other hand simulations are usually not realistic. We have done extremely realistic simulations of cardiac image sequences. This images, which underlying motion is perfecly controlled, can hardly be differnciated from real images delivered by commercial scanners.

The futur of this project will naturally concern 3D US-Tagging. The construction of matrix arrays that enable 3D ultrasound imaging is possible. The contrôle of the image formation with such arrays should permit 3D US-Tagging.
Also it should be possible to imagine applying aour technique to other organs or tissues as for example blood vessels. Indeed the formation of the plaque inside the arteries usually involves changes in the way the vessel border is moving during the cardiac cycle.

This project has lead to 1 journal article and 7 conference communications . Other articles are in the submission cycle or beeing written.

Context
This project investigates the detection of cardiovascular pathologies using a new ultrasound imaging method: UltraSound Tagging. It is well known by clinicians that the analysis of the motion and deformation of the heart muscle can contribute relevant information on the pathological state of the heart and be of great utility for diagnosis.
Although the techniques in echocardiography have greatly progressed and now provide access to high-quality images, currently there is no totally automatic imaging technique available to analyze these images and provide exact information about motion and deformation.
There are numerous studies in the literature on this subject. One of the most promising of them, published in 2005 by Ledesma Carbayo et al., is based on optical flow. It was shown that their technique could be used on clinical cases. However, according to the authors themselves, there is an important limitation inherent to the images.
We propose to study the complete acquisition and processing pipeline including the image formation.
Contributions
The main contribution of this study is situated in the image formation step. We would like to provide an image that facilitates 2D motion estimation and deformation of the heart. As in MRI, where tagging is used to superimpose patterns on the images in order to facilitate motion estimation, we would like to provide a beamformer dedicated to cardiac imaging done in sectorial geometry and the specific motion estimation method. Given our vast experience in this domain, we plan to work with transverse oscillation images used for blood flow and elastography and the phase-based motion estimation techniques. These methods were initially developed in linear geometry and they require substantial methodological work to be adapted to sectorial geometry.
We will then propose a segmentation and structure tracking algorithm that includes a motion estimation prior. The segmentation problem will be studied in the context of level-sets.
Finally, we will propose a set of visualization and interaction tools for the clinician so that the relevant information can be extracted for diagnosis.
Organization
This project is organized around four young scientists. This consortium combines expertise in methodology with a vast experience at all the steps of image processing (image formation, motion estimation, segmentation, structure tracking) and with medical proficiency with a clinician among the team members.
The project is split into two axes (methodological and experimental) and divided into five tasks. Tasks 1 and 2 are precursors to the development of the image formation technique, motion estimation, and structure tracking. Task 3 is a simulation validation task. Task 4 implements the method on a ultrasound research scanner. Task 5 is the experimental validation task divided into first a phantom validation and second a feasibility study in vivo.
Repercussions
This multidisciplinary project has the potential of becoming a real alternative to conventional echocardiography methods.
The health benefits are clear given that cardiovascular disease is one of the main causes of death in industrialized countries.
The economic potential of the project is extremely important, for both the company providing this method on its scanner, as it will be a strong commercial argument, and the CREATIS laboratory. This project could well be extended as a technology transfer to industry.