SOM-QE ANALYSIS: A biologically inspired technique to detect and track meaningful changes within image regions

Abstract

In practice, finding evidence for subtle changes in critical image regions through visual expert inspection of serial imaging data can be challenging. For scans taken at relatively short intervals, relevant changes may be too small to be noticed, yet diagnostically meaningful. The earliest such detection generates critical insights into potential risks, the fast it permits setting up early control mechanisms or strategies for clinical treatment. SOM-QE algorithm automatically detects subtle but significant changes in image time series providing information likely to be meaningful for experts. It is implemented in Python to analyze medical, geographic, or behavioral data.