Classification of amyotrophic lateral sclerosis by brain volume, connectivity, and network dynamics

Abstract

Emerging studies corroborate the importance of neuroimaging biomarkers and machine learning to improve diagnostic classification of amyotrophic lateral sclerosis (ALS). While most studies focus on structural data, recent studies assessing functional connectivity between brain regions by linear methods highlight the role of brain function. These studies have yet to be combined with brain structure and nonlinear functional features. We investigate the role of linear and nonlinear functional brain features, and the benefit of combining brain structure and function for ALS classification. ALS patients (N = 97) and healthy controls (N = 59) underwent structural and functional resting state magnetic resonance imaging. Based on key hubs of resting state networks, we defined three feature sets comprising brain volume, resting state functional connectivity (rsFC), as well as (nonlinear) resting state dynamics assessed via recurrent neural networks. Unimodal and multimodal random forest classifiers were built to classify ALS. Out‐of‐sample prediction errors were assessed via five‐fold cross‐validation. Unimodal classifiers achieved a classification accuracy of 56.35–61.66%. Multimodal classifiers outperformed unimodal classifiers achieving accuracies of 62.85–66.82%. Evaluating the ranking of individual features' importance scores across all classifiers revealed that rsFC features were most dominant in classification. While univariate analyses revealed reduced rsFC in ALS patients, functional features more generally indicated deficits in information integration across resting state brain networks in ALS. The present work undermines that combining brain structure and function provides an additional benefit to diagnostic classification, as indicated by multimodal classifiers, while emphasizing the importance of capturing both linear and nonlinear functional brain properties to identify discriminative biomarkers of ALS.

Segmental alterations of the corpus callosum in motor neuron disease: A DTI and texture analysis in 575 patients

“…An SVM-based analysis of six callosal parameters could achieve a separation of all MND patients (ALS patients including phenotypes) from controls with 84% sensitivity and 85% specificity. This is, with respect to an AUC value of 0.91, an excellent result according to predefined criteria ( Mandrekar, 2010 ) in comparison to other applications of multiparametric MRI classifiers to ALS (review in Thome et al, 2022 ) and supports the use of CC as a neuroimaging marker in ALS. A recent study ( Bede et al, 2022A ) used a perceptron model to discriminate ALS, UMN predominant, and LMN predominant cohorts and found that measures of the forceps minor (anterior CC / fibres of the genu of CC) discriminate these subtypes rather well, in the same line like the results of the current study; approaches of neural network classification based on neuroimaging features have also been reported ( Bede et al, 2022B ).…”

Functional networks in prolonged disorders of consciousness

“…Previous studies have mainly applied fMRI to the baseline consciousness assessment and brain function exploration in prolonged DoC (Crone et al, 2014;Weng et al, 2017;Zhang et al, 2018), providing insights into the neural mechanisms of brain networks that have not been fully understood so far. In addition to brain injury, functional connectivity and network integrity are also disturbed to varying degrees in aging (Malagurski et al, 2020;Patil et al, 2021) and neurodegenerative disorders including mild cognitive impairment, Alzheimer's disease, Parkinson's disease, and amyotrophic lateral sclerosis (Zhu et al, 2021;Miao et al, 2022;Thome et al, 2022;Zhao et al, 2022). Of which, the DMN is highly vulnerable.…”

Aberrant Multimodal Connectivity Pattern Involved in Default Mode Network and Limbic Network in Amyotrophic Lateral Sclerosis

“…This is consistent with previous studies that showed widespread structural (but not functional) damage in ALS patients compared with HCs [36,39]. Incorporating the significant differences in SC-FC coupling as features in the SVM model resulted in classification performance with similar or even higher accuracy (72.09-81.40%), sensitivity (72.00-88.00%), and specificity (61.11-94.44%) for distinguishing ALS patients from HCs compared to that obtained in previous studies using other neu-roimaging features [42][43][44][45]. Our findings show that SC-FC coupling may be a valuable neuroimaging marker for ALS.…”