|
| Summary document | |
Version
française
Today, practicing doctors have many medical images at their disposal:
X-rays, tomography, magnetic resonance imaging, isotopic imaging, ultrasound,
histologic images, video images and so on.
How can they make use of all this information for diagnosis, therapeutics
and surgery?
 |
© INRIA |
Imaging is sharply growing sector that has now become an integral
part of medical practice. New technology makes it possible to access increasingly
precise views of the human body and its functioning. For example, in vivo
molecular imaging reveals tissue activity at the microscopic level and
provides very rich information on the pathological functioning of systems.
Classical techniques are also making progress toward increased resolution.
All these images now make up an information of such richness and complexity
that specific software is needed to analyze and make use of it.
The Epidaure research team is entirely devoted to the topic of imaging
for medicine. Their work concerns the integration, interpretation and
utilization of the whole spectrum of medical images stemming from different
media, the processing and storage of corresponding information and their
use in guiding robotics.
Epidaure explores many research areas, especially the compilation of anatomical
and functional atlases, the modeling of physiological systems and the
identification of important parameters in diagnosis and therapeutics.
This work is of interest to many partners :
At the same time, Epidaure continuously hosts doctoral candidates and
medical interns that enhance their training within the time while actively
participating in research development.
The project in brief
 |
© INRIA |
ISA is a team specializing in 3D computer
graphics and computer vision, fields that require building 3D models based
on actual images. Medicine provides a natural application of their work,
especially in the field of "augmented reality" based on medical images. The
goal of augmented reality is to show users a view of the object under observation
(a patient, an organ) that is enhanced by supplementary data or a simulated
evolution.
The reconstruction and merging techniques used make it possible to recompose
a complete image from several sources, angiography and MRI for example, or
from a series of images taken a few minutes apart.
Interventional neurosurgery is one of the fields of application of this work,
in partnership with the regional teaching hospital of Nancy and General Electric
(GEMSE). In 2003, the research concerned the construction of a precise image
of arterio venous malformations (AVMs) for purposes of radiotherapy treatment.
Indeed, radiotherapy tools are now capable of treating complex shapes. It
is thus important to spot and precisely describe such shapes based on medical
images. The team is also working on nephrological scanner analysis.
ISA is a joint project of CNRS, INRIA, INPL and the Henri Poincaré Nancy
1 and Nancy 2 Universities. The project also applies its work in earth sciences
and plasma physics.
The team has international collaborations with partners in the ARIS European
project as well as through associated McGill-ISA teams. Industrial relations
are strong General Electric Medical Systems, SGI, CEA, CIRTES and SGDL. Such
collaborations made it possible to develop such software applications as
Gocad, Candela and Graphite. The work of the team enabled former project
members to
launch three startups, Earth Decision Sciences (initially T-Surf), Neoxy
and VSP Technologie.
Could increasingly precise medical images constitute a reliable tool to aid in diagnostic, therapeutic and surgical decisions?
 |
© INRIA |
Today, medical images
are interpreted by doctors who are faced with increasingly numerous and
complex information sources. As in other sectors, information management
technology should make it possible to develop decision aid tools that take
into account all the available data and acquired experience. The value
of these tools is foremost to improve treatments by providing doctors with
support. They could also contribute to the traceability of decisions.
Such are the main stakes in the research of the VISAGES team. The team
chose to focus its work on applications concerning pathologies of the neck,
the head and the brain.
An image is however a source of complex information in itself. This is
why one of VISAGES' research fields is the processing and merging of images
from multiple sources as well as the segmentation and analysis of the information
they contain.
Surgical operation guiding and remote ultrasound are some of the clinical
applications of this work, especially in the treatment of epilepsy, multiple
sclerosis and Parkinson's disease.
The team was built around a common resolve to integrate work starting from
upstream research down to clinical tests in the hospital. It gathers together
research scientists from INRIA, CNRS, the University of Rennes and the
Rennes Hospital.
VISAGES developed another research direction, information sharing (images
and image processing algorithms) to improve patient care through professional
networking including radiologists, neurologists and hospital departments.
The research team is working on the development of a "Neurobase" by
setting up a system infrastructure not only to exchange data, but also
image processing tools. This project in partnership with the IFR 49, INSERM,
IRISA and INRIA is supported by the Ministry of Research.
The project in brief
How to conduct complex medical examinations from a distance with the same
precision and efficacy?
Can automatic control and image modeling open the way for remote ultrasound?
 |
© INRIA |
What a dream remote medicine would be for astronauts in space or sailors
in a submarine! However, the stakes are just as high in the day to day
practice of terrestrial medicine. Certain pathologies demand very specialized
examinations that are only performed by experts who practice in the city
or in large hospital centers. Now, moving a patient to an examination center
is a heavy and costly process. It would be much more comfortable for the
patient and more cost effective for the community if he or she could be
examined right on the spot, in the generalist's office or in a local hospital
center.
If remote control for certain devices is becoming widespread, one of the
problems still facing research scientists is the delay between a complex
command and its remote implementation. This is one of the research directions
of the LAGADIC team. The team applies its work on visual servo control
to the development of remote ultrasound. The team's scientists think it
is possible to do better than remotely handle the ultrasound probe. One
possibility would be to have the examination robot be servo controlled
by the image it captures, so that the expert could concentrate on the analysis.
For example, the expert could ask the device to turn around the organ.
The robot would then compute the trajectory of the probe making sure it
stays well in contact with the patient's skin. The number of commands issued
by the operator would thus be diminished. The LAGADIC team (meaning "little
eye" in the Breton language) completes the device with a camera that
supplies a few of the examination field.
This approach requires complex methodological problems to be solved: the
identification of "efficient" visual information, which entails
its modeling, the association of images supplied by the ultrasound probe
with those supplied by the camera to maintain a good skin/probe interaction,
the generation of 3D images from ultrasound images (in slices), the coupling
between an off center vision of the examination field and a stress sensor
for remotely operated examinations, to name a few.
LAGADIC is testing its work in the framework of the 3D ultrasound acquisition
robotic cell, a robotized platform it shares with the Visages team.
Medicine is only one of the applications of Lagadic work in the fields
of visual servo control and robotics. Others are underwater robotics, food
processing, mobile robotics in transports, etc.
