Upper motor neuron burden measurement in motor neuron diseases: Does one scale fit all?

VoxelHop: Successive Subspace Learning for ALS Disease Classification Using Structural MRI

Deep learning has great potential for accurate detection and classification of diseases with medical imaging data, but the performance is often limited by the number of training datasets and memory requirements. In addition, many deep learning models are considered a "black-box," thereby often limiting their adoption in clinical applications. To address this, we present a successive subspace learning model, termed VoxelHop, for accurate classification of Amyotrophic Lateral Sclerosis (ALS) using T2-weighted structural MRI data. Compared with popular convolutional neural network (CNN) architectures, VoxelHop has modular and transparent structures with fewer parameters without any backpropagation, so it is well-suited to small dataset size and 3D imaging data. Our VoxelHop has four key components, including (1) sequential expansion of near-to-far neighborhood for multi-channel 3D data; (2) subspace approximation for unsupervised dimension reduction; (3) label-assisted regression for supervised dimension reduction; and (4) concatenation of features and classification between controls and patients. Our experimental results demonstrate that our framework using a total of 20 controls and 26 patients achieves an accuracy of 93.48 % and an AUC score of 0.9394 in differentiating patients from controls, even with a relatively small number of datasets, showing its robustness and effectiveness. Our thorough evaluations also show its validity and superiority to the state-of-the-art 3D CNN classification approaches. Our framework can easily be generalized to other classification tasks using different imaging modalities.

Rate-dependent depression is impaired in amyotrophic lateral sclerosis

OBJECTIVE

We investigated rate-dependent depression (RDD) of the Hoffman reflex (H-reflex) in patients with amyotrophic lateral sclerosis (ALS), a degenerative disease with ventral horn involvement.

PATIENTS AND METHODS

In this case-control study, we enrolled 27 patients with ALS and 30 matched healthy control subjects. Clinical and electrophysiological assessments, as well as RDD in response to various stimulation frequencies (0.5 Hz, 1 Hz, 3 Hz and 5 Hz), were compared between groups. Multiple clinical and electrophysiological factors were also explored to determine any underlying associations with RDD.

RESULTS

The ALS group showed a significant loss of RDD across all frequencies compared to the control group, most notably following 1 Hz stimulation (19.1 ± 20.3 vs. 34.0 ± 13.7%, p = 0.003). Among factors that might influence RDD, the enlargement of the motor unit potential (MUP) showed a significant relationship with RDD following multifactor analysis of variance (p = 0.007) and Pearson correlation analysis (ρ =  - 0.70, p < 0.001), while various upper motor neuron manifestations were not correlated with RDD values (p > 0.05).

CONCLUSION

We report a loss of RDD in patients with ALS. The strong correlation detected between the RDD deficit and increased MUP suggests that RDD is a sensitive indicator of underlying spinal disinhibition in ALS.

TRIAL REGISTRATION

ChiCTR2000038848, 10/7/2020 (retrospectively registered), http://www.chictr.org.cn/ .

Comparison of Two Clinical Upper Motor Neuron Burden Rating Scales in ALS Using Quantitative Brain Imaging

Several clinical upper motor neuron burden scales (UMNSs) variably measure brain dysfunction in amyotrophic lateral sclerosis (ALS). Here, we compare relationship of two widely used clinical UMNSs in ALS (Penn and MGH UMNSs) with (a) neuroimaging markers of brain dysfunction and (b) neurological impairment status using the gold-standard functional measure, the revised ALS Functional Rating Scale (ALSFRS-R). MGH UMNS measures hyperreflexia alone, and Penn UMNS measures hyperreflexia, spasticity, and pseudobulbar affect. Twenty-eight ALS participants underwent both Penn and MGH UMNSs, at a matching time-point as a simultaneous [¹¹C]PBR28 positron emission tomography (PBR28-PET)/Magnetic Resonance scan and ALSFRS-R. The two UMNSs were compared for localization and strength of association with neuroimaging markers of: (a) neuroinflammation, PBR28-PET and MR Spectroscopy metabolites (myo-inositol and choline) and (b) corticospinal axonal loss, fractional anisotropy (FA), and MR Spectroscopy metabolite (N-acetylaspartate). Among clinical UMN manifestations, segmental hyperreflexia, spasticity, and pseudobulbar affect occurred in 100, 43, and 18% ALS participants, respectively. Pseudobulbar affect did not map to any specific brain regional dysfunction, while hyperreflexia and spasticity subdomains significantly correlated and colocalized neurobiological changes to corticospinal pathways on whole brain voxel-wise analyses. Both UMNS total scores showed significant and similar strength of association with (a) neuroimaging changes (PBR28-PET, FA, MR Spectroscopy metabolites) in primary motor cortices and (b) severity of functional decline (ALSFRS-R). Hyperreflexia is the most frequent clinical UMN manifestation and correlates best with UMN molecular imaging changes in ALS. Among Penn UMNS's subdomains, hyperreflexia carries the weight of association with neuroimaging markers of biological changes in ALS. A clinical UMN scale comprising hyperreflexia items alone is clinically relevant and sufficient to predict the highest yield of molecular neuroimaging abnormalities in ALS.