1- , Department of Biomedical Engineering, SR.C., Islamic Azad University, Tehran, Iran
2- , Department of Biomedical Engineering, SR.C., Islamic Azad University, Tehran, Iran ,a-sheikhani@srbiau.ac.ir
2- , Department of Biomedical Engineering, SR.C., Islamic Azad University, Tehran, Iran ,
Abstract: (90 Views)
Introduction: Differential diagnosis of Alzheimer’s disease and frontotemporal dementia remains a challenge for insurance organizations due to overlapping clinical manifestations, costly imaging and laboratory examinations. Early diagnosis refers to identifying disease at its initial stages, whereas differential diagnosis focuses on distinguishing diseases with similar clinical manifestations, which is the objective of this study. According to the World Alzheimer Report 2021, cost of dementia diagnosis depending on the type and extent of specialized assessments may be increase. For example, in Quebec, Canada, the cost of dementia diagnosis in 2021 ranged from 160 to 4,058$ for assessments involving MRI, cerebrospinal fluid analysis, and PET, highlighting the importance of developing more accessible and cost-effective diagnostic approaches.
Methods: A dataset from the neurology team of AHEPA University Hospital in Thessaloniki, comprising 36 individuals with Alzheimer’s disease, 23 with frontotemporal dementia, and 29 healthy individuals, was used. EEG signal features were extracted using the Network in Network architecture and integrated with clinical features for classification. Model performance was evaluated using 10-fold cross-validation.
Results: The model differentiated Alzheimer’s disease, frontotemporal dementia, and healthy individuals with an accuracy of 78.12%.
Conclusion: Compared with some previous studies based on three-class End-to-End deep learning networks, which reported accuracies below 70% and, in some cases, as low as 54%, the model demonstrated superior performance. findings indicate that hybrid deep learning can serve as a clinical decision-support tool alongside clinical assessments to facilitate differential diagnosis and enable more targeted use of specialized and costly diagnostic methods in basic insurance organizations.
Methods: A dataset from the neurology team of AHEPA University Hospital in Thessaloniki, comprising 36 individuals with Alzheimer’s disease, 23 with frontotemporal dementia, and 29 healthy individuals, was used. EEG signal features were extracted using the Network in Network architecture and integrated with clinical features for classification. Model performance was evaluated using 10-fold cross-validation.
Results: The model differentiated Alzheimer’s disease, frontotemporal dementia, and healthy individuals with an accuracy of 78.12%.
Conclusion: Compared with some previous studies based on three-class End-to-End deep learning networks, which reported accuracies below 70% and, in some cases, as low as 54%, the model demonstrated superior performance. findings indicate that hybrid deep learning can serve as a clinical decision-support tool alongside clinical assessments to facilitate differential diagnosis and enable more targeted use of specialized and costly diagnostic methods in basic insurance organizations.
Keywords: Alzheimer's disease, Frontotemporal dementia, Deep learning, Cost management, health insurance organizations
Type of Study: Research |
Subject:
Special
Received: 2026/09/14 | Revised: 2026/09/14 | Accepted: 2026/09/1
Received: 2026/09/14 | Revised: 2026/09/14 | Accepted: 2026/09/1
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
