University of Illinois Urbana-Champaign

Using machine learning techniques and brain MRI scans for detection of Alzheimer’s disease

Liu, Yuyang; Mazumdar, Suvodeep; Bath, Peter

Content Files
liu_et_al_abstract.docm
liu_et_al_poster.pdf

Permalink

https://hdl.handle.net/2142/113739

Description

Title
Using machine learning techniques and brain MRI scans for detection of Alzheimer’s disease
Author(s)
  • Liu, Yuyang
  • Mazumdar, Suvodeep
  • Bath, Peter
Issue Date
2022-02-28
Keyword(s)
  • Alzheimer’s Disease
  • Machine Learning
  • MRI
Date of Ingest
2022-04-22T20:16:34Z
Abstract
Dementia is a clinical syndrome characterized by cognitive and behavioral impairment: it mostly affects people who are aged 65 years and over. Dementia results from several diseases, of which Alzheimer’s disease (AD) accounts for up to 80% of all dementia diagnoses. Magnetic Resonance Imaging (MRI) is one of the most widely used methods to diagnose AD but due to low efficiency of manual analysis, machine learning algorithms have been developed to diagnose AD using medical imaging data. In this study, unsupervised learning strategies were used to cluster the two diagnostic status, a healthy status called cognitively normal (CN), and AD, using brain structural MRI scans. First, we detected the abnormal regions between CN and AD using two-sample t-tests, and then employed an unsupervised learning neural network to extract features from brain MRI images. In the final stage, unsupervised learning (clustering) was implemented to discriminate between CN and AD data based on the extracted features. The approach was tested on 429 individuals from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) who had baseline brain structural MRI scans: 231 CN and 198 AD. In the study, we found that the abnormal regions around the hippocampus were indicated based on two-sample t-test (p<0.0001), and the proposed methods using the abnormal regions yield the clustering results for CN vs. AD (accuracy=0.8163, specificity=0.7863, sensitivity=0.8436, and precision=0.8411 [mean values based on 10 runs]).
Publisher
iSchools
Type of Resource
Other
Genre of Resource
Conference Poster
Language
eng
Permalink
http://hdl.handle.net/2142/113739
Copyright and License Information
Copyright 2022 is held by Yuyang Liu, Suvodeep Mazumdar, and Peter Bath. Copyright permissions, when appropriate, must be obtained directly from the authors.

Owning Collections

iConference 2022 Posters PRIMARY