Parkinson’s, alzheimer’s disease diagnosis using fdg-pet images with neural networks

Yathiraj, Alok (2020) Parkinson’s, alzheimer’s disease diagnosis using fdg-pet images with neural networks. Master's Internship Report, Biomedical Engineering.

Preview	Text [mBME]_[2020]_[Alok Yathiraj].pdf Download (1MB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (150kB)

Abstract

Parkinson’s (PD) and Alzheimer’s (AD) disease are two of the most common neurological diseases worldwide. The diagnosis of these two diseases is especially hard. The diagnosis for both these diseases are subjective, making them unreliable as results could vary clinic to clinic. With each of these diseases having different treatment plans it is imperative to diagnose them accurately. It has been found that some areas of the brain degenerate and others become more active for each of these neurological diseases, this change in activity can be visualised using [ 18 F]-fluoro-deoxyglucose positron emission tomography (FDG-PET) which depicts the metabolic activity of the brain. It has also been found that using Scaled Subprofile Modelling/Principal Component Analysis (SSM/PCA) on the FDG-PET images helps by converting the data into a space where disease specific patterns of covariation in the brain can be seen easily. This makes it easier to distinguish between different patterns seen in the FDG-PET images. The goal of this paper is to analyse the effectiveness of using feed feed forward neural networks and the influence of using two different activation functions to classify AD from PD patients, using FDG-PET images that have been pre-processed using SSM/PCA as the input features. Additionally, performance of the network to classify AD from healthy controls (HC) and PD from HC is also tested with the parameters of each network chosen such that the highest accuracy can be achieved.

Item Type:	Thesis (Master's Internship Report)
Supervisor name:	Greuter, M.J.W.
Degree programme:	Biomedical Engineering
Thesis type:	Master's Internship Report
Language:	English
Date Deposited:	01 Sep 2020 13:50
Last Modified:	01 Sep 2020 13:50
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/23332

Actions (login required)

View Item