Multiparametric Microstructural MRI and Machine Learning Classification Yields High Diagnostic Accuracy in Amyotrophic Lateral Sclerosis: Proof of Concept.

Thomas D. Kocar,Anna Behler,Albert C. Ludolph,Hans-Peter Müller,Jan Kassubek

doi:10.3389/fneur.2021.745475

Abstract

The potential of multiparametric quantitative neuroimaging has been extensively discussed as a diagnostic tool in amyotrophic lateral sclerosis (ALS). In the past, the integration of multimodal, quantitative data into a useful diagnostic classifier was a major challenge. With recent advances in the field, machine learning in a data driven approach is a potential solution: neuroimaging biomarkers in ALS are mainly observed in the cerebral microstructure, with diffusion tensor imaging (DTI) and texture analysis as promising approaches. We set out to combine these neuroimaging markers as age-corrected features in a machine learning model with a cohort of 502 subjects, divided into 404 patients with ALS and 98 healthy controls. We calculated a linear support vector classifier (SVC) which is a very robust model and then verified the results with a multilayer perceptron (MLP)/neural network. Both classifiers were able to separate ALS patients from controls with receiver operating characteristic (ROC) curves showing an area under the curve (AUC) of 0.87–0.88 (“good”) for the SVC and 0.88–0.91 (“good” to “excellent”) for the MLP. Among the coefficients of the SVC, texture data contributed the most to a correct classification. We consider these results as a proof of concept that demonstrated the power of machine learning in the application of multiparametric quantitative neuroimaging data to ALS.

Highlights

Amyotrophic lateral sclerosis (ALS) is a clinically and genetically heterogeneous, multidomain neurodegenerative syndrome of motor and extra-motor systems with multiple different clinical subphenotypes and alterations in several brain regions, most prominently the corticospinal tract (CST) [1]
The data were collected from the magnetic resonance imaging (MRI) data archive of the Dept of Neurology, University of Ulm, Germany and included 1.5 T imaging data sets that contained a high-resolution T1 weighted imaging (T1w) sequence and a diffusion tensor imaging (DTI) sequence with at least 39 gradients
The leaveone-out cross-validation (LOOCV) analysis confirmed these results with 80% sensitivity, 80% specificity, and an area under the curve (AUC) of 87% in the receiver operating characteristic (ROC) analysis

Summary

Introduction

Amyotrophic lateral sclerosis (ALS) is a clinically and genetically heterogeneous, multidomain neurodegenerative syndrome of motor and extra-motor systems with multiple different clinical subphenotypes and alterations in several brain regions, most prominently the corticospinal tract (CST) [1]. The combination of texture analysis with traditional diffusion metrics makes a multiparametric microstructural assessment possible that might enhance diagnostic accuracy in machine learning classifiers. A retrospective data analysis with T1w and DTI data was conducted To this end, we extracted diffusion metrics of the most important tracts in ALS as well as texture data from the motor segment of the corpus callosum. We extracted diffusion metrics of the most important tracts in ALS as well as texture data from the motor segment of the corpus callosum We combined these data in a linear support vector classifier (SVC) which is a robust model that can give feedback on feature importance, with the prospect of providing a proof of concept model with high accuracy and unraveling underlying patterns that could help to build future classifiers. As a proof of concept study, we primarily focused on the general feasibility of the approach, not the optimization of the ML algorithm

Methods

Results

Conclusion