Continuous Microglial Activation Score

Table of Contents

Abstract

Presentation

01. Introduction 02. Microglial Classification 03. Machine Learning Model 04. Activation Score 05. Alpha-Synuclein Mouse Model 06. Microglia Density 07. Microglia Activation 08. Activation vs. Density 09. Mouse Model Activation Score 10. Clustering 11. Summary

https://opt003stagmediafiles.blob.core.windows.net/image/f3f2b37ffc1d4f368050b29a17405d5a

Cells showing progressively higher activation score (from 0 to 1) from left to right. These cells show progressive changes in their morphology, supporting the use of this score as a metric of activation.

https://opt003stagmediafiles.blob.core.windows.net/image/fdf65f4b8fd54078a21649096e16becf

An example of how the model arrives at a decision using the Shapley (SHAP) values, and how the value of each features contribute toward an activation score.

In this example (cell shown in inset), the large soma, high number of branches, and low soma circularity contribute the most towards a high activation score. The value of each parameter (normalized to the average) is shown in gray and the positive SHAP values (shown in red) tend towards the activated state.

For example, low soma circularity (lower than average in this case) contributes towards activation. The SHAP values for each feature are summed to obtain the overall SHAP value for that particular cell.

--:--

To more precisely represent the continuous process of microglia activation, our model also outputs an activation score ranging from 0 to 1, where 1 represents the most activated state and 0.5 the threshold for the discrete classification. The activation scores are obtained by using a combination of the morphological features from the model. The continuous activation score can be used for finer quantification of the state of the microglia rather than the discrete classification.