CHLOE: A Software Tool for Automatic Novelty Detection in Microscopy Image Datasets

Abstract

The recent advancements in automated microscopy and information systems allow the acquisition and storage of massive datasets of microscopy images. Here we describe CHLOE, a software tool for automatic extraction of novelty in microscopy image datasets. The tool is based on a comprehensive set of numerical image content descriptors reflecting image morphology, and can be used in combination with ROI detection and segmentation tools such as ITK. The rich feature set allows automatic detection of repetitive outlier images that are visually different from the common images in the dataset. The code and software are publicly available for free download at http://vfacstaff.ltu.edu/lshamir/downloads/chloe.

Keywords

• Image analysis
• Microscopy
• Outlier detection.