Artificial intelligence–driven early detection of neurodegenerative diseases using multimodal neuroimaging data

artificial intelligence; deep learning; convolutional neural network; multimodal neuroimaging; MRI; fMRI; PET; early detection; neurodegenerative diseases; multimodal fusion

Khushvakova Nilufar Jurakulovna1,
Samarkand State Medical University, Professor,
Doctor of Medical Sciences, Head of the
Department of Otorhinolaryngology No. 1.
Samarkand, Uzbekistan.
e-mail: nilumedlor@mail.ru
ORCID:0009-00006717

Rustamov Mardon Rustamovich. 2
Professor, Doctor of Medical Sciences,
Department Nº1 - Pediatrics and Neonatology of
the Samarkand State Medical University.
rustamovmardon212@gmail.com
ORCID: 00000002-3573-6304

Khazratov Utkir 3
MD, PhD, Associate Professor, Department of
Internal Medicine Propaedeutics, Abu Ali ibn
Sina Bukhara State Medical Institute
https://orcid.org/0000-0003-0378-6355

Muzaffar Zokirov4
Associate professor, PhD, Fergana Medical
Institute of Public Health, Fergana city, Uzbekistan
muzaffarzokirov91@mail.ru
Orcid: 0009-0009-2916-2613

Ganiev Abdukamol5
PhD, associate Professor, Department of
Traumatology and Orthopedics, Tashkent State
Medical University, Tashkent .
E-mail:abdukamolganiev0112@gmail.com,
orcid.org/https://orcid.org/0000-0001-5787-5199

Vali janova Muattar6
Teacher of anatomy, Department of Medical
Fundamental Sciences, Namangan branch of
Tashkent, International University
Kimyovalijanovamuattar@gmail.com
Orcid: ttps://orcid.org/000900031867728X

Received-21-05-2026
Revised-22-06-2026
Accepted-27-06-2026

European Journal of Prosthodontics and Restorative Dentistry (2026) 34(4S), 150–155

Artificial intelligence–driven early detection of neurodegenerative diseases using multimodal neuroimaging data

Background: Neurodegenerative diseases such as Alzheimer’s and Parkinson’s disease develop gradually, with neuropathological changes often preceding clinical symptoms by years. Early, accurate detection is critical for timely intervention and clinical trial stratification. Recent advances in artificial intelligence (AI) have enabled the extraction of subtle and multimodal imaging biomarkers that are difficult to discern through conventional radiological assessment. Objective: This study developed a deep learning–based multimodal fusion framework for the early detection of neurodegenerative diseases using magnetic resonance imaging (MRI), functional MRI (fMRI), and positron emission tomography (PET) data. Methods: Data from 1,020 participants (aged 55–85 years) were sourced from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Parkinson’s Progression Markers Initiative (PPMI). Each imaging modality underwent standardized preprocessing (bias correction, normalization, co- registration). A convolutional neural network (CNN) architecture with modality- specific feature extractors and an attention- based fusion layer was trained using five- fold cross- validation. Model performance was benchmarked against unimodal CNNs and gradient- boosted ensemble classifiers. Results: The multimodal AI model achieved an overall accuracy of 93.4%, sensitivity = 91.6%, and specificity = 95.1%, substantially outperforming unimodal MRI (85.8%) and PET (83.2%) networks. Feature saliency analysis highlighted hippocampal and posterior cingulate changes as key determinants in early Alzheimer’s detection, whereas basal ganglia connectivity patterns were predictive of prodromal Parkinson’s. Conclusion: The proposed AI- driven multimodal integration framework significantly enhances early diagnostic precision for neurodegenerative diseases by exploiting cross- modal neuroimaging synergies. These findings establish a foundation for clinical translation of deep learning pipelines in precision neurology and risk stratification for early therapeutic interventions.

•••••••••••••••••••••••••••••••• ejprd.org- Published by Riset Publishing Services LLC.

EJPRD

Copyright © 2026 by Riset Publishing Services LLC