A Representation of Changes of Images and its Application for Developmental Biolology

TL;DR

This paper introduces a method to represent and analyze time-series image data, particularly gene expression data in developmental biology, using function fitting and machine learning for classification and disease diagnosis.

Contribution

It proposes a novel representation technique for temporal biological data that captures intrinsic properties and enables classification and diagnosis.

Findings

Effective classification of developmental gene expression data

Potential for disease diagnosis based on data fluctuations

Method applicable to various time-varying biological data

Abstract

In this paper, we consider a series of events observed at spaced time intervals and present a method of representation of the series. To explain an idea, by dealing with a set of gene expression data, which could be obtained from developmental biology, the procedures are sketched with comments in some details. We mean representation by choosing a proper function, which fits well with observed data of a series, and turning its characteristics into numbers, which extract the intrinsic properties of fluctuating data. With help of a machine learning techniques, this method will give a classification of developmental biological data as well as any varying data during a certain period and the classification can be applied for diagnosis of a disease.