Automated and unbiased classification of motor neuron phenotypes with single cell resolution in ALS tissue.

Histopathological analysis of tissue sections is an invaluable resource in neurodegeneration research. Importantly, cell-to-cell variation in both the presence and severity of a given phenotype is however a key limitation of this approach, reducing the signal to noise ratio and leaving unresolved the potential of single-cell scoring for a given disease attribute. Here, we developed an image processing pipeline for automated identification and profiling of motor neurons (MNs) in amyotrophic lateral sclerosis (ALS) pathological tissue sections. This approach enabled unbiased analysis of hundreds of cells, from which hundreds of features were readily extracted. Next by testing different machine learning methods, we automated the identification of phenotypically distinct MN subpopulations in VCP- and SOD1-mutant transgenic mice, revealing common aberrant phenotypes in cellular shape. Additionally we established scoring metrics to rank cells and tissue samples for both disease probability and severity. Finally, by adapting this methodology to human post-mortem tissue analysis, we validated our core finding that morphological descriptors strongly discriminate ALS from control healthy tissue at the single cell level. In summary, we show that combining automated image processing with machine learning methods substantially improves the speed and reliability of identifying phenotypically diverse MN populations. Determining disease presence, severity and unbiased phenotypes at single cell resolution might prove transformational in our understanding of ALS and neurodegenerative diseases more broadly.