Activity Number:
|
221
- Contributed Poster Presentations: Section on Statistics in Imaging
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Section on Statistics in Imaging
|
Abstract #313303
|
|
Title:
|
Quantification of Pixel-Wise Noise in Clinical Positron Emission Tomography (PET) Images
|
Author(s):
|
Ran Ren* and Jian Huang and Finbarr O'Sullivan and Tian Mou and Kevin O'Regan
|
Companies:
|
University College Cork and University College Cork and University College Cork and Karolinska Institutet and Cork University Hospital
|
Keywords:
|
PET/CT;
Re-binning;
Attenuation;
Gamma
|
Abstract:
|
Clinical PET/CT-FDG scanning is an important diagnostic tool applied in a number of prevalent cancers. Recently, our group have developed a Gamma model for modelling noise in iteratively reconstructed PET images and validated it on uniform phantom data (Tian et al 2017). In this study we develop a novel method for estimation of local Gamma model parameters associated with non-uniform sources. The estimation method is based on combination of local homogeneity and attenuation effects and consists of two steps: 1. For each pixel, we estimate local Gamma parameters by fitting gamma distribution to a local VOI centered at that pixel; 2. We refine estimates based on the attenuation map (Kueng et al 2017). The techniques is evaluated using 31 lung cancer patients data from clinical whole-body scanning with reconstructed consecutive 1-minute time-frames. The analysis finds that there is strong agreement between the directly estimated using consecutive 1-minute time-frames and constructed by the proposed approach involving only the single 1-minute frame information.
|
Authors who are presenting talks have a * after their name.