TY - JOUR
T1 - Classification of Drosophila embryonic developmental stage range based on gene expression pattern images.
AU - Ye, Jieping
AU - Chen, Jianhui
AU - Li, Q.
AU - Kumar, Sudhir
PY - 2006
Y1 - 2006
N2 - The genetic analysis of spatial patterns of gene expression relies on the direct visualization of the presence or absence of gene products (mRNA or protein) at a given developmental stage (time) of a developing animal. The raw data produced by these experiments include images of the Drosophila embryos showing a particular gene expression pattern revealed by a gene-specific probe. The identification of genes showing spatial and temporal overlaps in their expression patterns is fundamentally important to formulating and testing gene interaction hypotheses. Comparison of expression patterns is most biologically meaningful when images from a similar time point (developmental stage range) are compared. In this paper, we propose a computational system for automatic developmental stage classification by image analysis. This classification system uses image textural properties at a sub-block level across developmental stages as distinguishing features. Gabor filters are applied to extract features of image sub-blocks. Robust implementations of Linear Discriminant Analysis (LDA) are employed to extract the most discriminant features for the classification. Experiments on a collection of 2705 expression pattern images from early stages show that the proposed system significantly outperforms previously reported results in terms of classification accuracy, which shows high promise of the proposed system in reducing the time taken by biologists to assign the embryo stage range.
AB - The genetic analysis of spatial patterns of gene expression relies on the direct visualization of the presence or absence of gene products (mRNA or protein) at a given developmental stage (time) of a developing animal. The raw data produced by these experiments include images of the Drosophila embryos showing a particular gene expression pattern revealed by a gene-specific probe. The identification of genes showing spatial and temporal overlaps in their expression patterns is fundamentally important to formulating and testing gene interaction hypotheses. Comparison of expression patterns is most biologically meaningful when images from a similar time point (developmental stage range) are compared. In this paper, we propose a computational system for automatic developmental stage classification by image analysis. This classification system uses image textural properties at a sub-block level across developmental stages as distinguishing features. Gabor filters are applied to extract features of image sub-blocks. Robust implementations of Linear Discriminant Analysis (LDA) are employed to extract the most discriminant features for the classification. Experiments on a collection of 2705 expression pattern images from early stages show that the proposed system significantly outperforms previously reported results in terms of classification accuracy, which shows high promise of the proposed system in reducing the time taken by biologists to assign the embryo stage range.
UR - http://www.scopus.com/inward/record.url?scp=34250833265&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=34250833265&partnerID=8YFLogxK
U2 - 10.1142/18609475730038
DO - 10.1142/18609475730038
M3 - Article
C2 - 17369647
AN - SCOPUS:34250833265
SN - 1752-7791
SP - 293
EP - 298
JO - Computational systems bioinformatics / Life Sciences Society. Computational Systems Bioinformatics Conference
JF - Computational systems bioinformatics / Life Sciences Society. Computational Systems Bioinformatics Conference
ER -