Texture classification of MR images of the brain in ALS using M-CoHOG: A multi-center study

SF2Former: Amyotrophic lateral sclerosis identification from multi-center MRI data using spatial and frequency fusion transformer

Amyotrophic Lateral Sclerosis (ALS) is a complex neurodegenerative disorder characterized by motor neuron degeneration. Significant research has begun to establish brain magnetic resonance imaging (MRI) as a potential biomarker to diagnose and monitor the state of the disease. Deep learning has emerged as a prominent class of machine learning algorithms in computer vision and has shown successful applications in various medical image analysis tasks. However, deep learning methods applied to neuroimaging have not achieved superior performance in classifying ALS patients from healthy controls due to insignificant structural changes correlated with pathological features. Thus, a critical challenge in deep models is to identify discriminative features from limited training data. To address this challenge, this study introduces a framework called SF2Former, which leverages the power of the vision transformer architecture to distinguish ALS subjects from the control group by exploiting the long-range relationships among image features. Additionally, spatial and frequency domain information is combined to enhance the network's performance, as MRI scans are initially captured in the frequency domain and then converted to the spatial domain. The proposed framework is trained using a series of consecutive coronal slices and utilizes pre-trained weights from ImageNet through transfer learning. Finally, a majority voting scheme is employed on the coronal slices of each subject to generate the final classification decision. The proposed architecture is extensively evaluated with multi-modal neuroimaging data (i.e., T1-weighted, R2*, FLAIR) using two well-organized versions of the Canadian ALS Neuroimaging Consortium (CALSNIC) multi-center datasets. The experimental results demonstrate the superiority of the proposed strategy in terms of classification accuracy compared to several popular deep learning-based techniques.

Phenotypic categorisation of individual subjects with motor neuron disease based on radiological disease burden patterns: A machine-learning approach

Motor neuron disease is an umbrella term encompassing a multitude of clinically heterogeneous phenotypes. The early and accurate categorisation of patients is hugely important, as MND phenotypes are associated with markedly different prognoses, progression rates, care needs and benefit from divergent management strategies. The categorisation of patients shortly after symptom onset is challenging, and often lengthy clinical monitoring is needed to assign patients to the appropriate phenotypic subgroup. In this study, a multi-class machine-learning strategy was implemented to classify 300 patients based on their radiological profile into diagnostic labels along the UMN-LMN spectrum. A comprehensive panel of cortical thickness measures, subcortical grey matter variables, and white matter integrity metrics were evaluated in a multilayer perceptron (MLP) model. Additional exploratory analyses were also carried out using discriminant function analyses (DFA). Excellent classification accuracy was achieved for amyotrophic lateral sclerosis in the testing cohort (93.7%) using the MLP model, but poor diagnostic accuracy was detected for primary lateral sclerosis (43.8%) and poliomyelitis survivors (60%). Feature importance analyses highlighted the relevance of white matter diffusivity metrics and the evaluation of cerebellar indices, cingulate measures and thalamic radiation variables to discriminate MND phenotypes. Our data suggest that radiological data from single patients may be meaningfully interpreted if large training data sets are available and the provision of diagnostic probability outcomes may be clinically useful in patients with short symptom duration. The computational interpretation of multimodal radiology datasets herald viable diagnostic, prognostic and clinical trial applications.

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

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.

Feature selection from magnetic resonance imaging data in ALS: a systematic review

Background:

With the advances in neuroimaging in amyotrophic lateral sclerosis (ALS), it has been speculated that multiparametric magnetic resonance imaging (MRI) is capable to contribute to early diagnosis. Machine learning (ML) can be regarded as the missing piece that allows for the useful integration of multiparametric MRI data into a diagnostic classifier. The major challenges in developing ML classifiers for ALS are limited data quantity and a suboptimal sample to feature ratio which can be addressed by sound feature selection.

Methods:

We conducted a systematic review to collect MRI biomarkers that could be used as features by searching the online database PubMed for entries in the recent 4 years that contained cross-sectional neuroimaging data of subjects with ALS and an adequate control group. In addition to the qualitative synthesis, a semi-quantitative analysis was conducted for each MRI modality that indicated which brain regions were most commonly reported.

Results:

Our search resulted in 151 studies with a total of 221 datasets. In summary, our findings highly resembled generally accepted neuropathological patterns of ALS, with degeneration of the motor cortex and the corticospinal tract, but also in frontal, temporal, and subcortical structures, consistent with the neuropathological four-stage model of the propagation of pTDP-43 in ALS.

Conclusions:

These insights are discussed with respect to their potential for MRI feature selection for future ML-based neuroimaging classifiers in ALS. The integration of multiparametric MRI including DTI, volumetric, and texture data using ML may be the best approach to generate a diagnostic neuroimaging tool for ALS.