Soft attention based DenseNet model for Parkinson's disease classification using SPECT images

ObjectiveDeep learning algorithms have long been involved in the diagnosis of severe neurological disorders that interfere with patients' everyday tasks, such as Parkinson's disease (PD). The most effective imaging modality for detecting the condition is DaTscan, a variety of single-photon...

Full description

Saved in:
Bibliographic Details
Main Authors: Thakur, Mahima, Kuresan, Harisudha, Dhanalakshmi, Samiappan, Lai, Khin Wee, Wu, Xiang
Format: Article
Published: Frontiers Media SA 2022
Subjects:
Online Access:http://eprints.um.edu.my/41614/
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Universiti Malaya
id my.um.eprints.41614
record_format eprints
spelling my.um.eprints.416142023-11-08T02:57:52Z http://eprints.um.edu.my/41614/ Soft attention based DenseNet model for Parkinson's disease classification using SPECT images Thakur, Mahima Kuresan, Harisudha Dhanalakshmi, Samiappan Lai, Khin Wee Wu, Xiang RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry ObjectiveDeep learning algorithms have long been involved in the diagnosis of severe neurological disorders that interfere with patients' everyday tasks, such as Parkinson's disease (PD). The most effective imaging modality for detecting the condition is DaTscan, a variety of single-photon emission computerized tomography (SPECT) imaging method. The goal is to create a convolutional neural network that can specifically identify the region of interest following feature extraction. MethodsThe study comprised a total of 1,390 DaTscan imaging groups with PD and normal classes. The architecture of DenseNet-121 is leveraged with a soft-attention block added before the final classification layer. For visually analyzing the region of interest (ROI) from the images after classification, Soft Attention Maps and feature map representation are used. OutcomesThe model obtains an overall accuracy of 99.2% and AUC-ROC score 99%. A sensitivity of 99.2%, specificity of 99.4% and f1-score of 99.1% is achieved that surpasses all prior research findings. Soft-attention map and feature map representation aid in highlighting the ROI, with a specific attention on the putamen and caudate regions. ConclusionWith the deep learning framework adopted, DaTscan images reveal the putamen and caudate areas of the brain, which aid in the distinguishing of normal and PD cohorts with high accuracy and sensitivity. Frontiers Media SA 2022-07-13 Article PeerReviewed Thakur, Mahima and Kuresan, Harisudha and Dhanalakshmi, Samiappan and Lai, Khin Wee and Wu, Xiang (2022) Soft attention based DenseNet model for Parkinson's disease classification using SPECT images. Frontiers in Aging Neuroscience, 14. ISSN 1663-4365, DOI https://doi.org/10.3389/fnagi.2022.908143 <https://doi.org/10.3389/fnagi.2022.908143>. 10.3389/fnagi.2022.908143
institution Universiti Malaya
building UM Library
collection Institutional Repository
continent Asia
country Malaysia
content_provider Universiti Malaya
content_source UM Research Repository
url_provider http://eprints.um.edu.my/
topic RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry
spellingShingle RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry
Thakur, Mahima
Kuresan, Harisudha
Dhanalakshmi, Samiappan
Lai, Khin Wee
Wu, Xiang
Soft attention based DenseNet model for Parkinson's disease classification using SPECT images
description ObjectiveDeep learning algorithms have long been involved in the diagnosis of severe neurological disorders that interfere with patients' everyday tasks, such as Parkinson's disease (PD). The most effective imaging modality for detecting the condition is DaTscan, a variety of single-photon emission computerized tomography (SPECT) imaging method. The goal is to create a convolutional neural network that can specifically identify the region of interest following feature extraction. MethodsThe study comprised a total of 1,390 DaTscan imaging groups with PD and normal classes. The architecture of DenseNet-121 is leveraged with a soft-attention block added before the final classification layer. For visually analyzing the region of interest (ROI) from the images after classification, Soft Attention Maps and feature map representation are used. OutcomesThe model obtains an overall accuracy of 99.2% and AUC-ROC score 99%. A sensitivity of 99.2%, specificity of 99.4% and f1-score of 99.1% is achieved that surpasses all prior research findings. Soft-attention map and feature map representation aid in highlighting the ROI, with a specific attention on the putamen and caudate regions. ConclusionWith the deep learning framework adopted, DaTscan images reveal the putamen and caudate areas of the brain, which aid in the distinguishing of normal and PD cohorts with high accuracy and sensitivity.
format Article
author Thakur, Mahima
Kuresan, Harisudha
Dhanalakshmi, Samiappan
Lai, Khin Wee
Wu, Xiang
author_facet Thakur, Mahima
Kuresan, Harisudha
Dhanalakshmi, Samiappan
Lai, Khin Wee
Wu, Xiang
author_sort Thakur, Mahima
title Soft attention based DenseNet model for Parkinson's disease classification using SPECT images
title_short Soft attention based DenseNet model for Parkinson's disease classification using SPECT images
title_full Soft attention based DenseNet model for Parkinson's disease classification using SPECT images
title_fullStr Soft attention based DenseNet model for Parkinson's disease classification using SPECT images
title_full_unstemmed Soft attention based DenseNet model for Parkinson's disease classification using SPECT images
title_sort soft attention based densenet model for parkinson's disease classification using spect images
publisher Frontiers Media SA
publishDate 2022
url http://eprints.um.edu.my/41614/
_version_ 1783876728711544832