Quantitative assessment of alpha-synuclein pathology with AI
Parkinson’s disease (PD) is the second most common neurodegenerative disorder after Alzheimer’s. PD affects around 10 million people worldwide. α-synuclein is a neuronal protein that is linked genetically and neuropathologically to PD.
Researchers at Lundbeck trained deep learning neural networks to build AI models for performing an objective, quantitative assessment of α-synuclein pathology and to generate a spatiotemporal map of pathology spread in mouse brains.
Overview of study:
Traditionally neuronal cells have been counted from samples by stereological methods
With Aiforia Create the researchers trained a deep learning AI model to quantify α-synuclein positive neurons and thus the total area of pathology
They were then able to develop an anatomical map of relevant brain regions
The cell count produced by Aiforia correlated with manual methods
The first study was a test performed on pS129-α-synuclein positive staining in the Substantia Nigra and Amygdala
The manual cell counts (with ImageJ) were compared to the Aiforia quantification, with a significant Pearson correlation (p-value <0.0001)
Aiforia quantitation showed to be a fast and accurate method