Alzheimer's disease diagnosis in the metaverse
Jalal Safari Bazargani,Nasir Rahim,Abolghasem Sadeghi-Niaraki,Tamer Abuhmed,Houbing Song,Soo-Mi Choi
DOI: https://doi.org/10.1016/j.cmpb.2024.108348
2024-07-21
Abstract:Background and objective: The importance of early diagnosis of Alzheimer's Disease (AD) is by no means negligible because no cure has been recognized for it rather than some therapies only lowering the pace of progression. The research gap reveals information on the lack of an automatic non-invasive approach toward the diagnosis of AD, in particular with the help of Virtual Reality (VR) and Artificial Intelligence. Another perspective highlights that current VR studies fail to incorporate a comprehensive range of cognitive tests and consider design notes for elderlies, leading to unreliable results. Methods: This paper tried to design a VR environment suitable for older adults in which three cognitive assessments namely: ADAS-Cog, Montreal Cognitive Assessment (MoCA), and Mini Mental State Exam (MMSE), are implemented. Moreover, a 3DCNN-ML model was trained based on the corresponding cognitive tests and Magnetic Resonance Imaging (MRI) with different modalities using the Alzheimer's Disease Neuroimaging Initiative 2 (ADNI2) dataset and incorporated into the application to predict if the patient suffers from AD. Results: The model has undergone three experiments with different modalities (Cognitive Scores (CS), MRI images, and CS-MRI). As for the CS-MRI experiment, the trained model achieved 97%, 95%, 95%, 96%, and 94% in terms of precision, recall, F1-score, AUC, and accuracy respectively. The considered design notes were also assessed using a new proposed questionnaire based on existing ones in terms of user experience, user interface, mechanics, in-env assistance, and VR induced symptoms and effects. The designed VR system provided an acceptable level of user experience, with participants reporting an enjoyable and immersive experience. While there were areas for improvement, including graphics and sound quality, as well as comfort issues with prolonged HMD use, the user interface and mechanics of the system were generally well-received. Conclusions: The reported results state that our method's comprehensive analysis of 3D brain volumes and incorporation of cognitive scores enabled earlier detection of AD progression, potentially allowing for timely interventions and improved patient outcomes. The proposed integrated system provided us with promising insights for improvements in the diagnosis of AD using technologies.