qEEG as Biomarker for Alzheimer's Disease: Investigating Relative PSD Difference and Coherence Analysis.

Purpose: Electroencephalography (EEG) is a non-intrusive technique that provides comprehensive insights into the electrical activities of the brain's cerebral cortex. The brain signals obtained from EEGs can be used as a neuropsychological biomarker to detect different stages of Alzheimer's disease (AD) through quantitative EEG (qEEG) analysis. This paper investigates the difference in the abnormalities of resting state EEG (rEEG) signals between eyes-open (EOR) and eyes-closed (ECR) in AD by analyzing 19-scalp electrode EEG signals and making a comparison with healthy controls (HC). Participants and Methods: The rEEG data from 534 subjects (ages 40-90) consisting of 269 HC and 265 AD subjects in South Korea were used in this study. The qEEG for EOR and ECR states were performed separately for HC and AD subjects to measure the relative power spectrum density (PSD) and coherence with functional connectivity to evaluate abnormalities. The rEEG data were preprocessed and analyzed using EEGlab and Brainstorm toolboxes in MATLAB R2021a software, and statistical analyses were carried out using ANOVA. Results: Based on the Welch method, the relative PSD of the EEG EOR and ECR states difference in the AD group showed a significant increase in the delta frequency band of 19 EEG channels, particularly in the frontal, parietal, and temporal, than the HC groups. The delta power band on the source level was increased for the AD group and decreased for the HC group. In contrast, the source activities of alpha, beta, and gamma frequency bands were significantly reduced in the AD group, with a high decrease in the beta frequency band in all brain areas. Furthermore, the coherence of rEEG among different EEG electrodes was analyzed in the beta frequency band. It showed that pair-wise coherence between different brain areas in the AD group is remarkably increased in the ECR state and decreased after subtracting out the EOR state. Conclusion: The findings suggest that examining PSD and functional connectivity through coherence analysis could serve as a promising and comprehensive approach to differentiate individuals with AD from normal, which may benefit our understanding of the disease.

Classification of Aß State From Brain Amyloid PET Images Using Machine Learning Algorithm.

Background and Purpose: Analyzing brain amyloid positron emission tomography (PET) images to access the occurrence of ß-amyloid (Aß) deposition in Alzheimer's patients requires much time and effort from physicians, while the variation of each interpreter may differ. For these reasons, a machine learning model was developed using a convolutional neural network (CNN) as an objective decision to classify the Aß positive and Aß negative status from brain amyloid PET images. Methods: A total of 7,344 PET images of 144 subjects were used in this study. The 18F-florbetaben PET was administered to all participants, and the criteria for differentiating Aß positive and Aß negative state was based on brain amyloid plaque load score (BAPL) that depended on the visual assessment of PET images by the physicians. We applied the CNN algorithm trained in batches of 51 PET images per subject directory from 2 classes: Aß positive and Aß negative states, based on the BAPL scores. Results: The binary classification of the model average performance matrices was evaluated after 40 epochs of three trials based on test datasets. The model accuracy for classifying Aß positivity and Aß negativity was (95.00±0.02) in the test dataset. The sensitivity and specificity were (96.00±0.02) and (94.00±0.02), respectively, with an area under the curve of (87.00±0.03). Conclusions: Based on this study, the designed CNN model has the potential to be used clinically to screen amyloid PET images.

CNN for a Regression Machine Learning Algorithm for Predicting Cognitive Impairment Using qEEG.

Purpose: Electroencephalogram (EEG) signals give detailed information on the electrical brain activities occurring in the cerebral cortex. They are used to study brain-related disorders such as mild cognitive impairment (MCI) and Alzheimer's disease (AD). Brain signals obtained using an EEG machine can be a neurophysiological biomarker for early diagnosis of dementia through quantitative EEG (qEEG) analysis. This paper proposes a machine learning methodology to detect MCI and AD from qEEG time-frequency (TF) images of the subjects in an eyes-closed resting state (ECR). Participants and Methods: The dataset consisted of 16,910 TF images from 890 subjects: 269 healthy controls (HC), 356 MCI, and 265 AD. First, EEG signals were transformed into TF images using a Fast Fourier Transform (FFT) containing different event-rated changes of frequency sub-bands preprocessed from the EEGlab toolbox in the MATLAB R2021a environment software. The preprocessed TF images were applied in a convolutional neural network (CNN) with adjusted parameters. For classification, the computed image features were concatenated with age data and went through the feed-forward neural network (FNN). Results: The trained models', HC vs MCI, HC vs AD, and HC vs CASE (MCI + AD), performance metrics were evaluated based on the test dataset of the subjects. The accuracy, sensitivity, and specificity were evaluated: HC vs MCI was 83%, 93%, and 73%, HC vs AD was 81%, 80%, and 83%, and HC vs CASE (MCI + AD) was 88%, 80%, and 90%, respectively. Conclusion: The proposed models trained with TF images and age can be used to assist clinicians as a biomarker in detecting cognitively impaired subjects at an early stage in clinical sectors.

A machine-learning algorithm for predicting brain age using Rey-Osterrieth complex figure tests of healthy participants.

OBJECTIVE: Neuropsychologists widely use the Rey-Osterrieth complex figure test (RCFT) as part of neuropsychological test batteries to evaluate cognitive function and assess constructional ability, with age being the most significant factor. Our study investigated a supervised machine learning (ML) algorithm to predict brain age gap using RCFT drawings from the healthy elderly community for early dementia detection. PARTICIPANTS AND METHODS: RCFT drawings from 1,970 healthy subjects (ages 45-90 years) were collected from the Korean elderly community. We recorded subject demographic information including: age, gender, and education level. We trained the ML model with RCFT copies, immediate recall, delayed recall, and education level of the healthy subjects using CNN regression algorithm from Keras (https://keras.io/) with the Tensorflow library. RESULTS: The performance was evaluated by the mean absolute error (MAE) and root mean squared error (RMSE) between the predicted age and the chronological age based on a test dataset of 300 healthy subjects. The CNN regression model achieved an MAE of 7.2 years in predicting the brain age gap of the subjects, with an RMSE of 8.9 years. CONCLUSION: The MAE and RMSE accuracies of the CNN regression model predicting the brain age gap showed the model could be a potential biomarker for individual brain aging and a cost-effective method for early dementia detection.

Prediction of East Asian Brain Age using Machine Learning Algorithms Trained With Community-based Healthy Brain MRI.

Background and Purpose: Magnetic resonance imaging (MRI) helps with brain development analysis and disease diagnosis. Brain volumes measured from different ages using MRI provides useful information in clinical evaluation and research. Therefore, we trained machine learning models that predict the brain age gap of healthy subjects in the East Asian population using T1 brain MRI volume images. Methods: In total, 154 T1-weighted MRIs of healthy subjects (55-83 years of age) were collected from an East Asian community. The information of age, gender, and education level was collected for each participant. The MRIs of the participants were preprocessed using FreeSurfer(https://surfer.nmr.mgh.harvard.edu/) to collect the brain volume data. We trained the models using different supervised machine learning regression algorithms from the scikit-learn (https://scikit-learn.org/) library. Results: The trained models comprised 19 features that had been reduced from 55 brain volume labels. The algorithm BayesianRidge (BR) achieved a mean absolute error (MAE) and r squared (R2) of 3 and 0.3 years, respectively, in predicting the age of the new subjects compared to other regression methods. The results of feature importance analysis showed that the right pallidum, white matter hypointensities on T1-MRI scans, and left hippocampus comprise some of the essential features in predicting brain age. Conclusions: The MAE and R2 accuracies of the BR model predicting brain age gap in the East Asian population showed that the model could reduce the dimensionality of neuroimaging data to provide a meaningful biomarker for individual brain aging.

Neuropsychological test using machine learning for cognitive impairment screening.

OBJECTIVES: Neuropsychological tests (NPTs) are widely used tools to evaluate cognitive functioning. The interpretation of these tests can be time-consuming and requires a specialized clinician. For this reason, we trained machine learning models that detect normal controls (NC), cognitive impairment (CI), and dementia among subjects. PATIENTS AND METHODS: A total number of 14,927 subject datasets were collected from the formal neuropsychological assessments Seoul Neuropsychological Screening Battery (SNSB) by well-qualified neuropsychologists. The dataset included 44 NPTs of SNSB, age, education level, and diagnosis of each participant. The dataset was preprocessed and classified according to three different classes NC, CI, and dementia. We trained machine-learning with a supervised machine learning classifier algorithm support vector machine (SVM) 30 times with classification from scikit-learn (https://scikit-learn.org/stable/) to distinguish the prediction accuracy, sensitivity, and specificity of the models; NC vs. CI, NC vs. dementia, and NC vs. CI vs. dementia. Confusion matrixes were plotted using the testing dataset for each model. RESULTS: The trained model's 30 times mean accuracies for predicting cognitive states were as follows; NC vs. CI model was 88.61 ± 1.44%, NC vs. dementia model was 97.74 ± 5.78%, and NC vs. CI vs. dementia model was 83.85 ± 4.33%. NC vs. dementia showed the highest accuracy, sensitivity, and specificity of 97.74 ± 5.78, 97.99 ± 5.78, and 96.08 ± 4.33% in predicting dementia among subjects, respectively. CONCLUSION: Based on the results, the SVM algorithm is more appropriate in training models on an imbalanced dataset for a good prediction accuracy compared to natural network and logistic regression algorithms. The NC vs. dementia machine-learning trained model with SVM based on NPTs SNSB dataset could assist neuropsychologists in classifying the cognitive function of subjects.

Digital trail making test-black and white: Normal vs MCI.

OBJECTIVE: Trail Making Test-Black and White (TMT-B&W) was developed to assess the cognition of patients with mild cognitive impairment (MCI), dementia, and Alzheimer's disease. Collection and analysis of test results have been limited due to scoring time and efforts required from both administrators and patients during and after taking the test. To increase efficiency and reducing scoring time, a computer version touchscreen-based digital trail making test-black and white (dTMT-B&W) was developed on Android and it was administered on MCI versus cognitively normal controls (NC) participants. The current study examines the sensitivity of newly developed computer version dTMT-B&W on NC and MCI subjects. METHOD: dTMT-B&W was developed using MIT app inventor software, a web-based integrated development environment (IDE) with the Android development tools that are used to build fully functional applications for smartphones and tablets. A total of 44 participants were included, comprised of 22 NC and 22 MCI. The dTMT-TMT-B&W was administered to all NC and MCI subjects. RESULT: dTMT-B&W was designed to be as consistent with the pen-paper TMT-B&W (ppTMT-B&W) where the application is a standalone installation. dTMT-B&W is divided into two parts (Part-A and Part-B), in which the subject attempts to connect black and white numbered circles sequentially as quickly as possible, while still maintaining accuracy. Similarly, the paper-based TMT-B&W requires the subject to connect black and white numbered circles in ascending order, except on a sheet of paper rather than a tablet. dTMT-B&W successfully distinguished NC from MCI subjects. CONCLUSION: dTMT-B&W is an Android application that was successfully developed to be as consistent as possible with the original pen-paper TMT-B&W to establish equal concurrent validity, with some improved features embedded into the design and dTMT-B&W revealed a significant correlation with frontal executive function and this can help in early diagnosing subjects with MCI among NC subjects.

Comparison of RCF Scoring System to Clinical Decision for the Rey Complex Figure Using Machine-Learning Algorithm.

BACKGROUND AND PURPOSE: Interpreting the Rey complex figure (RCF) requires a standard RCF scoring system and clinical decision by clinicians. The interpretation of RCF using clinical decision by clinicians might not be accurate in the diagnosing of mild cognitive impairment (MCI) or dementia patients in comparison with the RCF scoring system. For this reason, a machine-learning algorithm was used to demonstrate that scoring RCF using clinical decision is not as accurate as of the RCF scoring system in predicting MCI or mild dementia patients from normal subjects. METHODS: The RCF dataset consisted of 2,232 subjects with formal neuropsychological assessments. The RCF dataset was classified into 2 datasets. The first dataset was to compare normal vs. abnormal and the second dataset was to compare normal vs. MCI vs. mild dementia. Models were trained using a convolutional neural network for machine learning. Receiver operating characteristic curves were used to compare the sensitivity, specificity, and area under the curve (AUC) of models. RESULTS: The trained model's accuracy for predicting cognitive states was 96% with the first dataset (normal vs. abnormal) and 88% with the second dataset (normal vs. MCI vs. mild dementia). The model had a sensitivity of 85% for detecting abnormal with an AUC of 0.847 with the first dataset. It had a sensitivity of 78% for detecting MCI or mild dementia with an AUC of 0.778 with the second dataset. CONCLUSIONS: Based on this study, the RCF scoring system has the potential to present more accurate criteria than the clinical decision for distinguishing cognitive impairment among patients.