Recognizing Early Signs of Dementia - Utilizing [18F]-Fluorodeoxyglucose Positron Emission Tomography Brain Imaging Data and Machine Learning

Aug 4, 2026

Artificial intelligence and expanded neuroimaging datasets enable stronger analysis of brain changes, yet prior machine-learning studies often relied on small datasets and produced results that were hard to compare due to inconsistent PET processing. This master’s thesis delivers a reproducible and generalizable analysis of [18F]FDG PET images aimed at detecting brain metabolic changes before dementia. A standardized Magia pipeline and machine-learning feature extraction supported classification across diagnostic groups using PET quantification measures and AIVO-labeled diagnoses.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages69

Reading Time2 h 54 min

AccessFree to download

Contents of the thesis

Abbreviations

1 Introduction

1.1 Thesis Overview

1.2 Dementia and Memory Disorders

1.3 Different Types of Dementia and Neuroimaging

1.4 Neurodegeneration and Biomarkers

1.5 Brain and Glucose Metabolism

1.6 [18F]-FDG-PET and Magia

1.7 Statistical Methods in PET

1.8 Machine Learning Models

1.8.1 Boosting the Analysis with Algorithms

1.8.2 MLM Trained with a Medium-Sized Data Set

Recognizing early signs of dementia utilizing [18F]-fluorodeoxyglucose positron emission tomography brain imaging data and machine learning Molecular Systems Biology/Department Life Technologies Master's thesis Author: Veikka Savila 6.6.2025 Turku The originality of this thesis has been checked in accordance with the University of Turku quality assurance system using the Turnitin Originality Check service. Master's thesis Subject: Molecular Systems Biology Author: Veikka Savila Title: Recognizing early signs of dementia utilizing [18F]-fluorodeoxyglucose positron emission tomography brain imaging data and machine learning Supervisor(s): Doc. Marco Bucci, MSc Janne Isojärvi Number of pages: 69 pages Date: 6.6.2025 Artificial intelligence, powered by advanced machine learning models, has revolutionized scientific research. In parallel, available neuroimaging data has increased. However, the utilization of machine learning models to analyse the data has been limited to small datasets, producing only directional results. Furthermore, processing of the positron emission tomography (PET) imaging data has varied, generating results inapplicable for crosscomparison. This thesis aimed to provide a reproducible and generalizable analysis of [18F] -Fluorodeoxyglucose (FDG) PET images with focus on revealing underlying changes in brain metabolism preceding dementia. Magia, a standardized pipeline for analysis of the PET scans, was utilized, as well as machine learning models on the data acquired from Magia, for feature extraction and classification between diseased and healthy subjects. This thesis applied methods from previous studies in the fields of neuroimaging, neuroinformatics, and machine learning. The PET imaging data, with available memory problem diagnoses, was extracted from AIVO database of Turku PET Centre. The Magia pipeline for image analysis utilized three PET image quantification methods: 1) standardized uptake value, 2) fractional uptake rate, and 3) linear regression Patlak plot. The results of the three quantification methods in combination with the diagnostic data from AIVO were used in the training of three separate machine learning models. The machine learning models were used to classify between A) healthy controls, B) other memory problems, C) memory disorders, and D) other subjects. The produced prediction models were able to outperform an uninformed guess but performed worse than models in prior studies. Achieving robust prediction models proved challenging for example due to misplaced data. Furthermore, due to underrepresentation of subjects within groups during the training of the models, the prediction models exhibited poor sensitivity. Multiple points of improvement were characterized during the study. Further research in the topic is required. A robust prediction model relying on PET image derived data rather than the images themselves will decrease the required resources in the study of memory problems. Over and above that, the features extracted will contribute to a better understanding of the pathophysiology of neurodegeneration. Key words: Alzheimer’s disease, automated neuroimaging data analysis, brain scan, dementia,[ 18F] FDG PET brain imaging, machine learning, neurodegeneration, pipeline Contents of the thesis Abbreviations ......................................................................................................3 1 Introduction ......................................................................................................6 1.1 Thesis Overview ........................................................................................6 1.2 Dementia and Memory Disorders ..............................................................6 1.3 Different Types of Dementia and Neuroimaging ........................................7 1.3.1 Differentiation between Dementias......................................................7 1.3.2 Different Stages

FAQs

What is the main goal of the thesis on early dementia detection?

To provide a reproducible and generalizable machine-learning analysis of [18F]FDG PET images, focusing on revealing underlying brain metabolic changes preceding dementia.

Which toolchain and data sources are used for PET image processing?

The Magia standardized pipeline is used for PET scan analysis, and PET imaging data with memory problem diagnoses are extracted from the AIVO database of the Turku PET Centre.

How are machine-learning models trained and what classifications are produced?

Three PET quantification methods combined with AIVO diagnostic labels are used to train separate machine-learning models, which classify among healthy controls, other memory problems, memory disorders, and other subjects.

Aug 4, 2026

Upload by Genevieve