Fractional Hölder mean-based image segmentation for mouse behavior analysis in conditional place preference test

Conditioned place preference (CPP) test is a common protocol where animals (usually rodents) are used to study the rewarding effects of addictive drugs and potential substances that may alleviate drug dependence. Automated CPP systems lead to increased throughput and reduced biasness; however, due to cost of such systems, most laboratories opt for manual testing. In this study, we present a new method for automatic analysis of CPP videos through a fractional Hölder mean (FHM)-based method for segmenting mouse from the video frames, which can later be used for analyzing mouse movement and behavior (fractional segmentation method), and eventually understanding the effect of addictive drug. We show that FHM satisfies some basic properties, such as convergence criteria and contractive criteria, which are the most important properties in segmentation. The proposed FHM is based on local minima and maxima, which makes it robust to contrast variation that is found in the video frames, allowing it to successfully separate mouse from the background. The proposed segmentation results show that the use of fractional Hölder mean-based threshold leads to a better pixel accuracy of object segmentation compared to other standard existing methods.