Artificial Intelligence-Assisted Multimodal Data Fusion 
for Early Diagnosis of Alzheimer's Disease

Qi Yang

doi:10.70711/mhr.v2i9.8486

Artificial Intelligence-Assisted Multimodal Data Fusion for Early Diagnosis of Alzheimer's Disease

Qi Yang

Abstract

Objective: To explore the application value of AI-assisted multimodal data fusion technology in early Alzheimer's Disease (AD)
diagnosis. Methods: A total of 280 subjects were selected from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database, including
70 early AD patients, 70 mild cognitive impairment (MCI) patients and 140 healthy controls. Multimodal data included structural magnetic
resonance imaging (sMRI) data, positron emission tomography (PET) data (amyloid-? tracer), and cerebrospinal fluid (CSF) biomarkers
(A?42, tau, p-tau181). An AI model was constructed by combining convolutional neural network (CNN) (for extracting sMRI and PET image
features) and long short-term memory (LSTM) network (for mining biomarker data correlation). The model was trained and verified via 5-fold
cross-validation, and its diagnostic performance was compared with traditional single-modal methods (doctor's sMRI reading and single CSF
marker detection). Results: The AI multimodal fusion model showed accuracy, sensitivity, specificity and AUC of 89.2%, 87.1%, 90.7%
and 0.93 respectively in distinguishing early AD patients from healthy controls; in distinguishing early AD from MCI patients, the indicators were 82.5%, 80.0%, 84.3% and 0.88. All indicators were significantly higher than those of traditional single-modal methods (P<0.05).
Conclusion: AI-assisted multimodal data fusion technology can effectively integrate AD early brain structural changes (via neuroimaging) and
molecular changes (via biomarkers), improving the accuracy and reliability of early AD diagnosis with good clinical application prospects.

Keywords

Early Alzheimer's Disease; AI Early Warning Model; Multimodal Data Fusion; Early Diagnosis

Full Text:

PDF

Included Database

References

[1] Alzheimer's Association. 2023 Alzheimer's disease facts and figures[J]. Alzheimer's & Dementia, 2023, 19(3): 1370-1499.

[2] Jack CR Jr, Knopman DS, Jagust WJ, et al. NIA-AA research framework: toward a biological definition of Alzheimer's disease[J].

Alzheimer's & Dementia, 2018, 14(4): 535-562.

[3] Van der Flier WM, Scheltens P. Imaging in Alzheimer's disease[J]. Nature Reviews Neurology, 2018, 14(8): 461-472.

[4] Blennow K, Hampel H, Weiner M, et al. Cerebrospinal fluid and plasma biomarkers in Alzheimer disease[J]. Nature Reviews Neurology, 2019, 15(3): 145-163.

[5] Esteban M, Ourselin S, Cardoso MJ. Deep learning for brain MRI segmentation: a review[J]. Medical Image Analysis, 2017, 45: 61-80.

[6] Zhang Y, Wang J, Li X, et al. Multimodal deep learning for Alzheimer's disease diagnosis using MRI, PET, and CSF data[J]. Computational Biology and Chemistry, 2020, 88: 107432.

[7] ADNI. Alzheimer's Disease Neuroimaging Initiative (ADNI) Database[EB/OL]. https://adni.loni.usc.edu/, 2024.

[8] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409.1556,

2014.

[9] Hochreiter S, Schmidhuber J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.

DOI: http://dx.doi.org/10.70711/mhr.v2i9.8486

Refbacks

There are currently no refbacks.