Using Image Translation To Synthesize Amyloid Beta From Structural MRI

Date
2024-01-15
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Amyloid-beta and brain atrophy are known hallmarks of Alzheimer’s Disease (AD) and can be quantified with positron emission tomography (PET) and structural magnetic resonance imaging (MRI), respectively. PET uses radiotracers that bind to amyloid-beta, whereas MRI can measure brain morphology. PET scans have limitations including cost, invasiveness (involve injections and ionizing radiation exposure), and have limited accessibility, making PET not practical for screening early-onset AD. Conversely, MRI is a cheaper, less-invasive (free from ionizing radiation), and is more widely available, however, it cannot provide the necessary molecular information. There is a known relationship between amyloid-beta and brain atrophy. This thesis aims to synthesize amyloid-beta PET images from structural MRI using image translation, an advanced form of machine learning. The developed models have reported high-similarity metrics between the real and synthetic PET images and high-degree of accuracy in radiotracer quantification. The results are highly impactful as it enables amyloid-beta measurements form every MRI, for free!
Description
Keywords
MRI, PET, Generative Adversarial Networks
Citation
Vega Lara, F. (2024). Using image translation to synthesize amyloid beta from structural MRI (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.