Automatic band detection on pulsed-field gel electrophoresis images

Din Chang Tseng, You Ching Lee

Research output: Contribution to journalArticlepeer-review

3 Scopus citations

Abstract

Automatic band detection is important for molecular subtyping on gel images. Automatic band detection has been conducted for many years, but only a few incomplete systems have been proposed. Here, we propose a completely automatic band detection system for pulsed-field gel electrophoresis (PFGE) images. The proposed approach mainly comprises lane segmentation and band extraction. In lane segmentation, we characterized the structural features and the spatial distribution of bands in PFGE images to fit pairs of parallel lines. In band extraction, a polynomial fitting method was used to remove the uneven background in a lane; we then used local gradient information to split the stuck bands and extract all bands in every lane. We compared the results with existing semiautomatic methods based on 20 varied PFGE images. The proposed method was shown to be superior to previous methods in most cases; moreover, the proposed approach is a completely automatic processing method.

Original languageEnglish
Pages (from-to)145-155
Number of pages11
JournalPattern Analysis and Applications
Volume18
Issue number1
DOIs
StatePublished - Feb 2014

Keywords

  • Automatic detection
  • Image segmentation
  • Pulsed-field gel electrophoresis

Fingerprint

Dive into the research topics of 'Automatic band detection on pulsed-field gel electrophoresis images'. Together they form a unique fingerprint.

Cite this