Sociodemographic data and APOE-ε4 augmentation for MRI-based detection of amnestic mild cognitive impairment using deep learning systems

Machine Learning Classification of Patients with Amnestic Mild Cognitive Impairment and Non-Amnestic Mild Cognitive Impairment from Written Picture Description Tasks

Individuals with Mild Cognitive Impairment (MCI), a transitional stage between cognitively healthy aging and dementia, are characterized by subtle neurocognitive changes. Clinically, they can be grouped into two main variants, namely patients with amnestic MCI (aMCI) and non-amnestic MCI (naMCI). The distinction of the two variants is known to be clinically significant as they exhibit different progression rates to dementia. However, it has been particularly challenging to classify the two variants robustly. Recent research indicates that linguistic changes may manifest as one of the early indicators of pathology. Therefore, we focused on MCI's discourse-level writing samples in this study. We hypothesized that a written picture description task can provide information that can be used as an ecological, cost-effective classification system between the two variants. We included one hundred sixty-nine individuals diagnosed with either aMCI or naMCI who received neurophysiological evaluations in addition to a short, written picture description task. Natural Language Processing (NLP) and a BERT pre-trained language model were utilized to analyze the writing samples. We showed that the written picture description task provided 90% overall classification accuracy for the best classification models, which performed better than cognitive measures. Written discourses analyzed by AI models can automatically assess individuals with aMCI and naMCI and facilitate diagnosis, prognosis, therapy planning, and evaluation.

Systematic comparison of 3D Deep learning and classical machine learning explanations for Alzheimer's Disease detection

Black-box deep learning (DL) models trained for the early detection of Alzheimer's Disease (AD) often lack systematic model interpretation. This work computes the activated brain regions during DL and compares those with classical Machine Learning (ML) explanations. The architectures used for DL were 3D DenseNets, EfficientNets, and Squeeze-and-Excitation (SE) networks. The classical models include Random Forests (RFs), Support Vector Machines (SVMs), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting (LightGBM), Decision Trees (DTs), and Logistic Regression (LR). For explanations, SHapley Additive exPlanations (SHAP) values, Local Interpretable Model-agnostic Explanations (LIME), Gradient-weighted Class Activation Mapping (GradCAM), GradCAM++ and permutation-based feature importance were implemented. During interpretation, correlated features were consolidated into aspects. All models were trained on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. The validation includes internal and external validation on the Australian Imaging and Lifestyle flagship study of Ageing (AIBL) and the Open Access Series of Imaging Studies (OASIS). DL and ML models reached similar classification performances. Regarding the brain regions, both types focus on different regions. The ML models focus on the inferior and middle temporal gyri, and the hippocampus, and amygdala regions previously associated with AD. The DL models focus on a wider range of regions including the optical chiasm, the entorhinal cortices, the left and right vessels, and the 4th ventricle which were partially associated with AD. One explanation for the differences is the input features (textures vs. volumes). Both types show reasonable similarity to a ground truth Voxel-Based Morphometry (VBM) analysis. Slightly higher similarities were measured for ML models.

Classifying Alzheimer's Disease Neuropathology Using Clinical and MRI Measurements

Background

Computer-aided machine learning models are being actively developed with clinically available biomarkers to diagnose Alzheimer's disease (AD) in living persons. Despite considerable work with cross-sectional in vivo data, many models lack validation against postmortem AD neuropathological data.

Objective

Train machine learning models to classify the presence or absence of autopsy-confirmed severe AD neuropathology using clinically available features.

Methods

AD neuropathological status are assessed at postmortem for participants from the National Alzheimer's Coordinating Center (NACC). Clinically available features are utilized, including demographics, Apolipoprotein E(APOE) genotype, and cortical thicknesses derived from ante-mortem MRI scans encompassing AD meta regions of interest (meta-ROI). Both logistic regression and random forest models are trained to identify linearly and nonlinearly separable features between participants with the presence (N = 91, age-at-MRI = 73.6±9.24, 38 women) or absence (N = 53, age-at-MRI = 68.93±19.69, 24 women) of severe AD neuropathology. The trained models are further validated in an external data set against in vivo amyloid biomarkers derived from PET imaging (amyloid-positive: N = 71, age-at-MRI = 74.17±6.37, 26 women; amyloid-negative: N = 73, age-at-MRI = 71.59±6.80, 41 women).

Results

Our models achieve a cross-validation accuracy of 84.03% in classifying the presence or absence of severe AD neuropathology, and an external-validation accuracy of 70.14% in classifying in vivo amyloid positivity status.

Conclusions

Our models show that clinically accessible features, including APOE genotype and cortical thinning encompassing AD meta-ROIs, are able to classify both postmortem confirmed AD neuropathological status and in vivo amyloid status with reasonable accuracies. These results suggest the potential utility of AD meta-ROIs in determining AD neuropathological status in living persons.

Predictive modelling of brain disorders with magnetic resonance imaging: A systematic review of modelling practices, transparency, and interpretability in the use of convolutional neural networks

Brain disorders comprise several psychiatric and neurological disorders which can be characterized by impaired cognition, mood alteration, psychosis, depressive episodes, and neurodegeneration. Clinical diagnoses primarily rely on a combination of life history information and questionnaires, with a distinct lack of discriminative biomarkers in use for psychiatric disorders. Symptoms across brain conditions are associated with functional alterations of cognitive and emotional processes, which can correlate with anatomical variation; structural magnetic resonance imaging (MRI) data of the brain are therefore an important focus of research, particularly for predictive modelling. With the advent of large MRI data consortia (such as the Alzheimer's Disease Neuroimaging Initiative) facilitating a greater number of MRI-based classification studies, convolutional neural networks (CNNs)-deep learning models well suited to image processing tasks-have become increasingly popular for research into brain conditions. This has resulted in a myriad of studies reporting impressive predictive performances, demonstrating the potential clinical value of deep learning systems. However, methodologies can vary widely across studies, making them difficult to compare and/or reproduce, potentially limiting their clinical application. Here, we conduct a qualitative systematic literature review of 55 studies carrying out CNN-based predictive modelling of brain disorders using MRI data and evaluate them based on three principles-modelling practices, transparency, and interpretability. We propose several recommendations to enhance the potential for the integration of CNNs into clinical care.

DAFT: A universal module to interweave tabular data and 3D images in CNNs

Prior work on Alzheimer's Disease (AD) has demonstrated that convolutional neural networks (CNNs) can leverage the high-dimensional image information for diagnosing patients. Beside such data-driven approaches, many established biomarkers exist and are typically represented as tabular data, such as demographics, genetic alterations, or laboratory measurements from cerebrospinal fluid. However, little research has focused on the effective integration of tabular data into existing CNN architectures to improve patient diagnosis. We introduce the Dynamic Affine Feature Map Transform (DAFT), a general-purpose module for CNNs that incites or represses high-level concepts learned from a 3D image by conditioning feature maps of a convolutional layer on both a patient's image and tabular clinical information. This is achieved by using an auxiliary neural network that outputs a scaling factor and offset to dynamically apply an affine transformation to the feature maps of a convolutional layer. In our experiments on AD diagnosis and time-to-dementia prediction, we show that the DAFT is highly effective in combining 3D image and tabular information by achieving a mean balanced accuracy of 0.622 for diagnosis, and mean c-index of 0.748 for time-to-dementia prediction, thus outperforming all baseline methods. Finally, our extensive ablation study and empirical experiments reveal that the performance improvement due to the DAFT is robust with respect to many design choices.

Prediction of Alzheimer's Disease Progression Based on Magnetic Resonance Imaging

The neuroimaging method of multimodal magnetic resonance imaging (MRI) can identify the changes in brain structure and function caused by Alzheimer's disease (AD) at different stages, and it is a practical method to study the mechanism of AD progression. This paper reviews the studies of methods and biomarkers for predicting AD progression based on multimodal MRI. First, different approaches for predicting AD progression are analyzed and summarized, including machine learning, deep learning, regression, and other MRI analysis methods. Then, the effective biomarkers of AD progression under structural magnetic resonance imaging, diffusion tensor imaging, functional magnetic resonance imaging, and arterial spin labeling modes of MRI are summarized. It is believed that the brain changes shown on MRI may be related to the cognitive decline in different prodrome stages of AD, which is conducive to the further realization of early intervention and prevention of AD. Finally, the deficiencies of the existing studies are analyzed in terms of data set size, data heterogeneity, processing methods, and research depth. More importantly, future research directions are proposed, including enriching data sets, simplifying biomarkers, utilizing multimodal magnetic resonance, etc. In the future, the study of AD progression by multimodal MRI will still be a challenge but also a significant research hotspot.

Data analysis with Shapley values for automatic subject selection in Alzheimer's disease data sets using interpretable machine learning

BACKGROUND

For the recruitment and monitoring of subjects for therapy studies, it is important to predict whether mild cognitive impaired (MCI) subjects will prospectively develop Alzheimer's disease (AD). Machine learning (ML) is suitable to improve early AD prediction. The etiology of AD is heterogeneous, which leads to high variability in disease patterns. Further variability originates from multicentric study designs, varying acquisition protocols, and errors in the preprocessing of magnetic resonance imaging (MRI) scans. The high variability makes the differentiation between signal and noise difficult and may lead to overfitting. This article examines whether an automatic and fair data valuation method based on Shapley values can identify the most informative subjects to improve ML classification.

METHODS

An ML workflow was developed and trained for a subset of the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort. The validation was executed for an independent ADNI test set and for the Australian Imaging, Biomarker and Lifestyle Flagship Study of Ageing (AIBL) cohort. The workflow included volumetric MRI feature extraction, feature selection, sample selection using Data Shapley, random forest (RF), and eXtreme Gradient Boosting (XGBoost) for model training as well as Kernel SHapley Additive exPlanations (SHAP) values for model interpretation.

RESULTS

The RF models, which excluded 134 of the 467 training subjects based on their RF Data Shapley values, outperformed the base models that reached a mean accuracy of 62.64% by 5.76% (3.61 percentage points) for the independent ADNI test set. The XGBoost base models reached a mean accuracy of 60.00% for the AIBL data set. The exclusion of those 133 subjects with the smallest RF Data Shapley values could improve the classification accuracy by 2.98% (1.79 percentage points). The cutoff values were calculated using an independent validation set.

CONCLUSION

The Data Shapley method was able to improve the mean accuracies for the test sets. The most informative subjects were associated with the number of ApolipoproteinE ε4 (ApoE ε4) alleles, cognitive test results, and volumetric MRI measurements.