Leveraging Deep Learning for Enhanced Detection of Alzheimer’s Disease Through Morphometric Analysis of Brain Images

This study investigates the efficacy of tensor-based morphometry (TBM) in detecting Alzheimer’s Disease (AD) using deep learning techniques. The primary focus is on discerning the volumetric variations in brain tissues characteristic of AD, Mild Cognitive Impairment (MCI), and cognitively normal (CN...

Full description

Saved in:
Bibliographic Details
Published in:Traitement du signal 2023-08, Vol.40 (4), p.1355-1365
Main Authors: Çelebi, Selahattin Barış, Emiroğlu, Bülent Gürsel
Format: Article
Language:English
Subjects:
Accuracy
Algorithms
Alzheimer's disease
Brain
Brain cancer
Brain research
Classification
Datasets
Deep learning
Dementia
Diagnosis
Hippocampus
Image enhancement
Machine learning
Magnetic resonance imaging
Medical imaging
Neural networks
Neuroimaging
Patients
Physiology
Success
Tensors
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This study investigates the efficacy of tensor-based morphometry (TBM) in detecting Alzheimer’s Disease (AD) using deep learning techniques. The primary focus is on discerning the volumetric variations in brain tissues characteristic of AD, Mild Cognitive Impairment (MCI), and cognitively normal (CN) conditions. TBM, as a measure of minute local volume differences, is employed as the distinguishing feature. The results are juxtaposed with those obtained from machine-learning-based methods, trained using a variety of medical images. Three unique models were developed for this purpose. The first model, trained using medial slices of the brain (train: 1622; test: 406), displayed an accuracy of less than 50%. The second model utilized axial brain slices procured at 5-pixel intervals, encompassing the hippocampus and the temporal lobe (train: 1632; test: 406), and demonstrated a significantly improved accuracy of 93%. The third model, fine-tuned with small kernel sizes to better extract localized changes from the image data used in the second model, achieved an accuracy of 92%. The findings suggest that the application of TBM and deep learning to medial slices alone is insufficient for an accurate diagnosis of AD. However, employing TBM with deep learning techniques to slices covering the hippocampus and temporal lobe can potentially offer a highly accurate approach for early AD detection. Notably, the use of small filters to extract detailed features from TBM did not enhance the model's performance. This research underscores the potential of deep learning in advancing the field of AD detection and diagnosis, providing crucial insights into the future development of diagnostic tools.
ISSN:0765-0019
1958-5608
DOI:10.18280/ts.400405