DONNART, Me. Clara (2021) Proximal optimization for the PET Imaging Reconstruction PRE - Research Project, ENSTA.

[img]Archive (ZIP)
3979Kb

Abstract

The objective of this internship is twofold. Initially, the goal is the implementation of a site specialized in the inventory of proximal operators, a practical optimization tool within the framework of non diferentiable functions. It was then necessary to code these functions in Matlab and implement the site in HTML so that they appear and are accessible to the user. Second, we apply proximal operators to the inverse problems of PET medical imaging (photon-emitting topography), involving a projection operator and a Poisson noise that strongly degrades the observed data. The aim of the internship in this second part is to implement several resolution algorithms and compare their effciency in time (speed of the algorithm) and image quality compared to an original image. In principal this part has an educational aim but it could eventually help the work of Claire Rossignole.

Item Type:Thesis (PRE - Research Project)
Uncontrolled Keywords:Maximisation Minimisation (MM)
Subjects:Mathematics and Applications
Life Sciences and Engineering
ID Code:8430
Deposited By:clara Donnart
Deposited On:17 mars 2022 09:24
Dernière modification:17 mars 2022 09:24

Repository Staff Only: item control page