Summary:
Comparisons of two microscopy images can be accomplished in many different ways. This project activity presents a recommendation system for selecting microscopy image similarity metrics according to biological application requirements. The motivation stems from the fact that a choice of similarity metric depends on a task with requirements that can guide a computer system in selecting the most suitable metric. The suitability of an image similarity metric is modeled as the sensitivity and invariance of the metric to microscopy image content and to dynamic changes of the microscopy image content. In this project activity, we describe a mathematical and experimental design of an image similarity metric recommendation system by building sensitivity signatures, and by querying this reference database of sensitivity signatures to retrieve a similarity metric according to the biological requirements. We illustrate the prototype recommendation system based on synthetic and measured images and by considering spectral calibration and spatial registration applications.
Overview of the modeling framework.
Challenge Problem(s):
Many conclusions in the biomedical field are based on comparing measured and reference image observations. One of the fundamental components of these comparisons is a measure of similarity. Visual inspection and the human perception of similarity play a prominent role in biomedical research. However, advances in microscopy imaging and the corresponding growth of image data lead to an increasingly high demand for the automation of visual comparisons and for a transition from expert-applied image similarity to computer-applied image similarity. The basic question that arises in this scenario is: How to select the microscopy image similarity metric that meets requirements of biological tasks? Our main motivation is to address this question by taking application specific requirements of microscopy image processing tasks and recommending users an image similarity metric that is the most sensitive to the task requirements.
The goal of this work is to address the mapping between application specific requirements on image similarity metrics and the plethora of existing image similarity computations. Our objective is to provide support for finding similarity metrics that closely match application requirements on image comparisons. In order to achieve the goal, several challenges need to be addressed:
Our main approach toward a recommendation system for choosing image similarity metrics consists of