post-doctoral fellowship in ai deep learning for medical imaging
LaTIM
State University, Browse similar opportunities
en
State University, Browse similar opportunities
AI deep learning in medical imaging, application in adaptive radiation therapy
Scientific context
External beam radiation therapy (EBRT) is a key step of the reference treatment of pelvic cancers, the most common cancers in the male population, especially with prostate cancer but also rectum and bladder cancers. This radiation therapy consists in irradiated the tumor using multiple treatment fraction spread out on several weeks while sparing organs at risks. However, its efficiency is hampered by the large anatomical deformations occurring between the treatment fractions, while the treatment plan is precisely optimized according to a fixed anatomy. Adaptive radiation therapy consists in updating patient anatomy before each treatment fraction using Cone Beam CT (CBCT) image, fig 1. This imaging system is embedded on the external beam radiation therapy system allowing quickly acquire images of the patient. However, such imaging system provides poor contrast and poor image quality making difficult to update the shape of organs with accuracy. The main challenge of the adaptive radiation therapy is to automatically contour organs based on the daily CBCT and the initial CT and MR images of the patient (see fig 1). Recently AI, especially deep learning, shows promising results on image processing for improve the medical workflow in radiotherapy. The aim of this position is to explore new deep learning methods to resolve challenge on adaptive radiotherapy.
Job description and missions
Organ segmentation may be resolve following different strategy using deep learning since the organ’s contours are always performed on the initial CT, the one used to define the treatment planning.
For example, a network may learn how to segment organs on the CBCT using a data set that include the ground truth from CT and/or MR. Another example is performing a non-rigid registration between the initial CT and the daily CBCT to update the patient anatomy. Or use a cross-modality synthesis strategy using adversarial neural architectures (GAN), to generate a synthesis CT (sCT) from the CBCT, and then register simply the CT and sCT to update the anatomy.
The postdoc position will consist in investigating and developing new approaches in deep learning. Based on a literature study, the postdoc will be free to explore and propose different strategies (segmentation, registration, image synthesis) to improve adaptive radiation therapy workflow. Consequent data set are already available at the lab (partner with the local University Hospital) allowing to achieve well validation and publications in both domains, medical physics and image processing.
Profile
We look for a candidate with a PhD in computer sciences, image processing, or applied mathematics. Autonomy, open-mindedness and motivation, as well as good English speaking/writing skills, are also expected. Some experience in deep learning or in medical imaging is appreciated but not required. This position is a good opportunity to learn and master one of these topics.
Position context
The postdoc will join the INSERM UMR1101 Laboratory of Medical Information Processing (LaTIM, Brest, France). Our research group is composed of 20 peoples including PhD students and others postdocs. The future recruited postdoc will work in collaboration with different academic and hospital partners. The position will be for an initial duration of one year and could be renewable. Salary is depending on the candidate’s experience.
Choose the country you wish to travel to study for free, work or volunteer
