SGD-DABiLSTM based MRI Segmentation for Alzheimer's disease Detection

Alzheimer's disease (AD) is a neurodegenerative ailment that causes memory and cognitive skills to deteriorate over time. It is a terrible neurological ailment that causes memory and cognition impairments, as well as behavioural issues, neuropsychiatric disorders and impairment in everyday task...

Full description

Saved in:
Bibliographic Details
Published in2022 4th International Conference on Advances in Computing, Communication Control and Networking (ICAC3N) pp. 1163 - 1169
Main Authors V, Nisha A, Pallikonda Rajasekaran, M., Vishnuvarthanan, G., Arunprasath, T., Ramaraj, Kottamalai
Format Conference Proceeding
LanguageEnglish
Published IEEE 16.12.2022
Subjects
Alzheimer's disease
Alzheimer's disease (AD)
Deep Attention Bidirectional Long Short-Term Memory
Deep learning
Error analysis
Image segmentation
Magnetic resonance imaging
Magnetic Resonance Imaging (MRI)
Spatial resolution
Stochastic Gradient Descent Optimisation (SGDO)
Task analysis
Online AccessGet full text

Cover

More Information
Summary:Alzheimer's disease (AD) is a neurodegenerative ailment that causes memory and cognitive skills to deteriorate over time. It is a terrible neurological ailment that causes memory and cognition impairments, as well as behavioural issues, neuropsychiatric disorders and impairment in everyday tasks. AD is diagnosed as a cause of death in elderly people. It is also one of the most difficult diseases to diagnose, especially in the early stages, using standard manual approaches. Magnetic resonance imaging (MRI) is a popular tool for detecting neurodegenerative disorders because of its high spatial resolution. In this work a deep attention bidirectional long short-term memory (DABiLSTM) with stochastic gradient optimisation (SGDO) is utilised for the detection of AD. For pre-processing brain MRI images, the gaussian bilateral filter is used. The anomaly section of the obtained image is segmented using a BiLSTM based on deep attention. The system is optimised using stochastic gradient descent optimization (SGDO), which minimises the neutral network's error rate. This work is implemented using the MATLAB tool and the Alzheimer's Disease Neuroimaging Initiative 2 (ADNI2) dataset. When compared with current approaches, the proposed method obtained an accuracy of 94.63 % in the detection of AD.
DOI:10.1109/ICAC3N56670.2022.10074493