Annonce

Les commentaires sont clos.

Deep learning tools for selectIve internal radiation Therapy of hepatic tumours (DELINEATE) – registration and deep learning segmentation

14 Octobre 2021


Catégorie : Doctorant


Where: Imagerie et Vision Artificielle (ImViA) laboratory, Dijon, France

Specific unit: Functional and Molecular Imaging, Medical Image Processing (IFTIM), Dijon, France –
https://imvia.u-bourgogne.fr/en/laboratory/iftim-team
Start of the PhD: January 2022
Salary: about 1 500 € net per month
Team: Jean-Louis Alberini, MD, PhD; Arnaud Boucher, PhD; Olivier Chevallier, MD; Sarah Leclerc,
PhD; Fabrice Meriaudeau, PhD; Tien-Phong Pham, PhD student; Romain Popoff, PhD; Benoît Presles,
PhD; Jean-Marc Vrigneaud, PhD
 

Project description :

Liver cancer is the sixth most common cancer in the world but is the second most frequent cause of
cancer death in men and the sixth leading cause of cancer death in women. Among the different types
of liver cancer, hepatocellular carcinoma (HCC) can be treated by selective internal radiation therapy
(SIRT), which consists in injecting selectively into the hepatic arteries yttrium-90 (Y90) β-radiation
emitter microspheres. Prior to Y90 bead injection, several examinations must be performed. First, a
hepatic angiography is acquired to identify the extrahepatic vessels that must be prophylactically
embolized to preserve healthy organs, and a baseline contrast-enhanced magnetic resonance imaging
(ceMRI) scan is acquired to visualise the tumour. Then, a simulation of the treatment is performed by
injecting Technetium-99m macroaggregated albumin (Tc99m-MAA) as a surrogate for Y90 particles. A
pre-treatment dosimetry is performed by acquiring a single-photon emission computed tomography
(SPECT)/computed tomography (CT) scan which allows to obtain the Tc99m-MAA distribution of activity
and to calculate the Y90 activity to prescribe. Once calculated, the appropriate amount of Y90
microspheres is injected into the patient and a positron emission tomography (PET)/CT scan is acquired
to verify that the Y90 activity distribution is consistent with the simulation of the treatment and to perform
post-treatment dosimetry. To be able to compute a proper dosimetry, it is necessary beforehand to
perform registrations and delineate accurately the tumour and liver volumes. The aim of this PhD
is first to perform registrations between the baseline ceMRI and SPECT/CT images (pre-treatment
dosimetry), and between the baseline ceMRI and the PET/CT images (post-treatment dosimetry)
and then to bring state-of-the-art deep learning segmentation methods to SIRT to segment the
liver and tumour volumes. The objective of the registration work is not to develop a new registration
method, neither a new tool, but rather to validate/optimise on our data existing approaches already
implemented in software packages such as NiftyReg 1 or elastix 2 . The PhD student will be helped in this
task by a trainee. The objective of the segmentation work is to develop two supervised deep learning
segmentation algorithms. A first supervised deep learning segmentation method that takes advantage
of the MRI structural information only (baseline ceMRI) and a second supervised co-segmentation
method that takes advantages of both the SPECT/CT and the baseline ceMRI images.
The student will be located at the ImViA laboratory and will work in close relation with the medical teams
of the University Hospital François Mitterrand and the Georges François Leclerc Centre both in Dijon,
France.
 
Person specification :
 
Candidates must hold at least an upper second-class degree or equivalent qualifications in a relevant
subject area such computer science, biomedical engineering or applied mathematics to apply to this
interdisciplinary project with clinicians, IT specialists, and medical physicists. A master's degree in a
relevant discipline and additional research experience would be an advantage.
 
Application :
Applications (including a CV and covering letter outlining your motivation for the position) should be
sent to Benoît Presles (benoit.presles@u-bourgogne.fr)
 
Closing date: 30 November 2021