Interpretable deep learning as a means for decrypting disease signature in multiple sclerosis

Characterizing the Contribution of Dependent Features in XAI Methods

Explainable Artificial Intelligence (XAI) provides tools to help understanding how AI models work and reach a particular decision or outcome. It helps to increase the interpretability of models and makes them more trustworthy and transparent. In this context, many XAI methods have been proposed to make black-box and complex models more digestible from a human perspective. However, one of the main issues that XAI methods have to face especially when dealing with a high number of features is the presence of multicollinearity, which casts shadows on the robustness of the XAI outcomes, such as the ranking of informative features. Most of the current XAI methods either do not consider the collinearity or assume the features are independent which, in general, is not necessarily true. Here, we propose a simple, yet useful, proxy that modifies the outcome of any XAI feature ranking method allowing to account for the dependency among the features, and to reveal their impact on the outcome. The proposed method was applied to SHAP, as an example of XAI method which assume that the features are independent. For this purpose, several models were exploited for a well-known classification task (males versus females) using nine cardiac phenotypes extracted from cardiac magnetic resonance imaging as features. Principal component analysis and biological plausibility were employed to validate the proposed method. Our results showed that the proposed proxy could lead to a more robust list of informative features compared to the original SHAP in presence of collinearity.

A window into the mind-brain-body interplay: Development of diagnostic, prognostic biomarkers, and rehabilitation strategies in functional motor disorders

BACKGROUND AND AIMS

Functional motor disorders (FMD) present a prevalent, yet misunderstood spectrum of neurological conditions characterized by abnormal movements (i.e., functional limb weakness, tremor, dystonia, gait impairments), leading to substantial disability and diminished quality of life. Despite their high prevalence, FMD often face delayed diagnosis and inadequate treatment, resulting in significant social and economic burdens. The old concept of psychological factors as the primary cause (conversion disorder) has been abandoned due to the need for more evidence about their causal role. According to a predictive coding account, the emerging idea is that symptoms and disability may depend on dysfunctions of a specific neural system integrating interoception, exteroception, and motor control. Consequently, symptoms are construed as perceptions of the body's state. Besides the main pathophysiological features (abnormal attentional focus, beliefs/expectations, and sense of agency), the lived experience of symptoms and their resulting disability may depend on an altered integration at the neural level of interoception, exteroception, and motor control.

METHODS AND MATERIALS

Our proposal aims to elucidate the pathophysiological mechanisms of FMD through a three-stage research approach. Initially, a large cohort study will collect behavioral, neurophysiological, and MRI biomarkers from patients with FMD and healthy controls, employing eXplainable Artificial Intelligence (XAI) to develop a diagnostic algorithm. Subsequently, validation will occur using patients with organic motor disorders. Finally, the algorithm's prognostic value will be explored post-rehabilitation in one subgroup of patients with FMD.

RESULTS

Data collection for the present study started in May 2023, and by May 2025, data collection will conclude.

DISCUSSION

Our approach seeks to enhance early diagnosis and prognostication, improve FMD management, and reduce associated disability and socio-economic costs by identifying disease-specific biomarkers.

TRIAL REGISTRATION

This trial was registered in clinicaltrials.gov (NCT06328790).

XAI-Based Assessment of the AMURA Model for Detecting Amyloid-β and Tau Microstructural Signatures in Alzheimer's Disease

Brain microstructural changes already occur in the earliest phases of Alzheimer's disease (AD) as evidenced in diffusion magnetic resonance imaging (dMRI) literature. This study investigates the potential of the novel dMRI Apparent Measures Using Reduced Acquisitions (AMURA) as imaging markers for capturing such tissue modifications.Tract-based spatial statistics (TBSS) and support vector machines (SVMs) based on different measures were exploited to distinguish between amyloid-beta/tau negative (A[Formula: see text]-/tau-) and A[Formula: see text]+/tau+ or A[Formula: see text]+/tau- subjects. Moreover, eXplainable Artificial Intelligence (XAI) was used to highlight the most influential features in the SVMs classifications and to validate the results by seeing the explanations' recurrence across different methods.TBSS analysis revealed significant differences between A[Formula: see text]-/tau- and other groups in line with the literature. The best SVM classification performance reached an accuracy of 0.73 by using advanced measures compared to more standard ones. Moreover, the explainability analysis suggested the results' stability and the central role of the cingulum to show early sign of AD.By relying on SVM classification and XAI interpretation of the outcomes, AMURA indices can be considered viable markers for amyloid and tau pathology. Clinical impact: This pre-clinical research revealed AMURA indices as viable imaging markers for timely AD diagnosis by acquiring clinically feasible dMR images, with advantages compared to more invasive methods employed nowadays.

Predicting disease-related MRI patterns of multiple sclerosis through GAN-based image editing

INTRODUCTION

Multiple sclerosis (MS) is a complex neurodegenerative disorder that affects the brain and spinal cord. In this study, we applied a deep learning-based approach using the StyleGAN model to explore patterns related to MS and predict disease progression in magnetic resonance images (MRI).

METHODS

We trained the StyleGAN model unsupervised using T1-weighted GRE MR images and diffusion-based ADC maps of MS patients and healthy controls. We then used the trained model to resample MR images from real input data and modified them by manipulations in the latent space to simulate MS progression. We analyzed the resulting simulation-related patterns mimicking disease progression by comparing the intensity profiles of the original and manipulated images and determined the brain parenchymal fraction (BPF).

RESULTS

Our results show that MS progression can be simulated by manipulating MR images in the latent space, as evidenced by brain volume loss on both T1-weighted and ADC maps and increasing lesion extent on ADC maps.

CONCLUSION

Overall, this study demonstrates the potential of the StyleGAN model in medical imaging to study image markers and to shed more light on the relationship between brain atrophy and MS progression through corresponding manipulations in the latent space.

MERGE: A model for multi-input biomedical federated learning

Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data and blood test results, are routinely collected but rarely considered in DL pipelines. Nevertheless, DL requires large datasets that often must be pooled from different institutions, raising non-trivial privacy concerns. Federated learning (FL) is a cooperative learning paradigm that aims to address these issues by moving models instead of data across different institutions. Here, we present a federated multi-input architecture using images and tabular data as a methodology to enhance model performance while preserving data privacy. We evaluated it on two showcases: the prognosis of COVID-19 and patients' stratification in Alzheimer's disease, providing evidence of enhanced accuracy and F1 scores against single-input models and improved generalizability against non-federated models.

Artificial Intelligence and Multiple Sclerosis: Up-to-Date Review

Multiple sclerosis (MS) remains a challenging neurological disorder for the clinician in terms of diagnosis and management. The growing integration of AI-based algorithms in healthcare offers a golden opportunity for clinicians and patients with MS. AI models are based on statistical analyses of large quantities of data from patients including "demographics, genetics, clinical and radiological presentation." These approaches are promising in the quest for greater diagnostic accuracy, tailored management plans, and better prognostication of disease. The use of AI in multiple sclerosis represents a paradigm shift in disease management. With ongoing advancements in AI technologies and the increasing availability of large-scale datasets, the potential for further innovation is immense. As AI continues to evolve, its integration into clinical practice will play a vital role in improving diagnostics, optimizing treatment strategies, and enhancing patient outcomes for MS. This review is about conducting a literature review to identify relevant studies on AI applications in MS. Only peer-reviewed studies published in the last four years have been selected. Data related to AI techniques, advancements, and implications are extracted. Through data analysis, key themes and tendencies are identified. The review presents a cohesive synthesis of the current state of AI and MS, highlighting potential implications and new advancements.

Deciphering multiple sclerosis disability with deep learning attention maps on clinical MRI

The application of convolutional neural networks (CNNs) to MRI data has emerged as a promising approach to achieving unprecedented levels of accuracy when predicting the course of neurological conditions, including multiple sclerosis, by means of extracting image features not detectable through conventional methods. Additionally, the study of CNN-derived attention maps, which indicate the most relevant anatomical features for CNN-based decisions, has the potential to uncover key disease mechanisms leading to disability accumulation. From a cohort of patients prospectively followed up after a first demyelinating attack, we selected those with T1-weighted and T2-FLAIR brain MRI sequences available for image analysis and a clinical assessment performed within the following six months (N = 319). Patients were divided into two groups according to expanded disability status scale (EDSS) score: ≥3.0 and < 3.0. A 3D-CNN model predicted the class using whole-brain MRI scans as input. A comparison with a logistic regression (LR) model using volumetric measurements as explanatory variables and a validation of the CNN model on an independent dataset with similar characteristics (N = 440) were also performed. The layer-wise relevance propagation method was used to obtain individual attention maps. The CNN model achieved a mean accuracy of 79% and proved to be superior to the equivalent LR-model (77%). Additionally, the model was successfully validated in the independent external cohort without any re-training (accuracy = 71%). Attention-map analyses revealed the predominant role of frontotemporal cortex and cerebellum for CNN decisions, suggesting that the mechanisms leading to disability accrual exceed the mere presence of brain lesions or atrophy and probably involve how damage is distributed in the central nervous system.