A Novel Otsu Watershed based Method Applied for DNA Scalograms Segmentation

The recent progress in next-generation sequencing technology has produced a huge amount of exponentially-generated genomic data. The processing of the latter requires large storage spaces. This study focuses on the eukaryotic genomes of C. elegans DNA organized into domains of different structures and activities. It presents an efficient automatic model to identify the DNA sequences. The developed approach is applied in order to code these sequences in the form of an “image” scalogram and, then, extract the DNA characteristic motifs by employing segmentation techniques in order to obtain its genomic signature. For this purpose, scalograms are segmented using improved watershed. We propose here modified watershed because the classic algorithm is sensitive to noise and may result in over-segmentation or under-segmentation. This modified watershed is based on gradient transformation, open-closed reconstruction and distance transformation. It is very important to use the more appropriate threshold when segmenting images. The choice of this threshold is based on “maximum between-class variance method” (Otsu). Finally, to assess the robustness of the proposed approach and show that it provides better PSNR and MSE when compared with the classic watershed.