Lane detection and tracking in PCR gel electrophoresis images

Abstract

This article presents a new scheme that aims to track the center of and detect lanes without any human interventions as the first step of the automated tool to analyze DNA fingerprints represented in PCR gel electrophoresis images. Although several research results have been previously reported to track the centers of and detect the lanes using projection profiles, due to the curve of the lanes it was not completed yet. To resolve the problem, we estimated the average lane width using k-means clustering algorithm and conducted subsequent local image processing. In the subsequent local image processing, we partitioned an input image into small images and found local maxima (potential lane centers) on the vertical projection in each partitioned image. Then, the lanes were composed by connecting the local maxima. 38 PCR gel images including 1235 lanes were used to evaluate the performance of the proposed scheme. They were divided into two groups including 10 training images and 28 testing images. The proposed scheme finally achieved the performance of F-measure of 1.000 computed from precision of 0.998 and recall of 1.000. Experimental results have shown that the proposed scheme is able to track the center of and detect lanes without any human intervention and it may be used as an automated tool to help researchers to analysis PCR gel electrophoresis images.