Future of Alzheimer's detection: Advancing diagnostic accuracy through the integration of qEEG and artificial intelligence

• Published in: NeuroImage (Orlando, Fla.) Vol. 317; p. 121373
• Main Authors: Rezaei, Sahar, Asadirad, Farzan, Motamedi, Alireza, Kamran, Mohammadsadegh, Parsa, Farzaneh, Samimi, Haniyeh, Ghannadikhosh, Parna, Zahmatyar, Mahdi, Hosseinzadeh, Seyed Ali, Arabi, Hossein
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 15.08.2025; Elsevier Limited; Elsevier
• Subjects: Accuracy; Aged; Alzheimer Disease - diagnosis; Alzheimer Disease - physiopathology; Alzheimer's disease; Artificial Intelligence; Biomarkers; Brain research; Cognitive ability; Costs; Deep Learning; Dementia; Dementia disorders; Diagnosis; Disease; Early detection; EEG; Electroencephalography; Electroencephalography - methods; Electroencephalography - trends; Humans; Integration; Life expectancy; Machine learning; Memory; Neural networks; Neurodegeneration; Neurodegenerative diseases; Pathology; Population studies; Quantitative electroencephalography; Radiology/Diagnostic Imaging; Wavelet transforms; ADD; CNN; FNN; LDA; AUROC; KNN; SVM; NNet; Early detection; AUC; DLBPD; PDD; Quantitative electroencephalography; FFT; Nbayes; NADD; RBF; GC; Alzheimer's disease; MODWT; EC; ML; CWT; qEEG; AD; EO; IQR; LogReg; PCC; MCI; RF; NC; SCD; PE; LIME; Machine learning; bvFTD; SUVR; DLB; HC; Artificial intelligence; maximal overlap discrete wavelet transform; frontotemporal dementia; local interpretable model-agnostic explanations; neural network; permutation entropy; convolutional neural networks; normal controls; logistic regression; support vector machines; K-nearest neighbours; Alzheimer’s disease; interquartile Range; Fast Fourier Transform; probable DLB; radial basis function; standardized uptake value ratios; non-Alzheimer’s disease dementia; eye-closed; dementia with Lewy bodies; feed-forward neural network; random forest; continuous wavelet transform; eye-open; granger causality; linear discriminant analysis; healthy controls; Pearson correlation coefficient; area under the receiver operating characteristic curve; Parkinson's disease dementia; area under the curve; subjective cognitive decline; Naive Bayes; mild cognitive impairment; Alzheimer's disease dementia
• ISSN: 1053-8119; 1095-9572; 1095-9572
• DOI: 10.1016/j.neuroimage.2025.121373

•Quantitative electroencephalography (qEEG) combined with machine learning (ML) is a promising tool for early dementia detection. qEEG-ML shows high accuracy for early diagnosis and differentiation of Alzheimer’s disease (AD), mild cognitive impairment (MCI), and dementia with Lewy bodies (DLB).•qEEG-ML provides a cost-effective and portable alternative to MRI and PET scans. Multimodal integration like qEEG + MRI may enhance diagnostic accuracy. Combining qEEG with other biomarkers and deep learning may improve clinical adoption and precision.•Traditional ML techniques like linear discriminant analysis (LDA) can still achieve high diagnostic accuracy, whereas more complex models, such as convolutional neural networks (CNNs), often demand advanced equipment and specialized expertise. Small sample sizes limit generalizability. Diverse datasets are necessary to ensure fairness and generalizability. This comprehensive review examines the integration of Quantitative Electroencephalography (qEEG) and Artificial Intelligence (AI) in the detection and diagnosis of Alzheimer's Disease (AD). Through systematic analysis of 11 key studies across multiple international databases, we evaluated various AI architectures, including machine learning algorithms and deep learning networks, applied to qEEG data for AD detection. The review encompasses studies with diverse subject populations, ranging from 35 to 890 participants, with mean ages between 66.94 and 74.8 years. Results demonstrate that AI-enhanced qEEG analysis achieves remarkable diagnostic accuracy, with Linear Discriminant Analysis (LDA) reaching 93.18% accuracy and 97.92% Area Under Curve (AUC). Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs) also showed promising results, with some models achieving up to 100% sensitivity in specific classifications. The integration of multiple data types and advanced feature extraction methods significantly improved diagnostic precision. Geographic diversity in research origins, spanning from Asia to Europe and the Americas, provides robust cross-cultural validation of findings. However, challenges persist in data quality, computational resources, and standardization of methodologies. This review highlights the significant potential of AI-enhanced qEEG as a non-invasive, cost-effective tool for the diagnosis of AD in its prodromal and dementia stages, while also identifying areas requiring further research to optimize its clinical application. These findings suggest that AI-enhanced qEEG analysis could revolutionize AD diagnosis, enabling earlier intervention and improved patient outcomes.