Wavelet-based automated DNA sizing of fragments through AFM image processing

Abstract

Atomic Force Microscope (AFM) is a procedure to investigate the size of the DNA fragments. In this paper, an algorithm is presented to determine the DNA fragment size from images of DNA molecules. The previous automated approach that uses conventional filters fails to have multidirectional and multiscale properties. This automated approach implements an algorithm on multiscale shift invariant wavelet decomposition to recover the DNA fragments. In order to avoid erroneously evaluated fragments, this algorithm also includes thinning and threshold process on Euclidean norm of wavelet decomposition. The Euclidean norm of multilevel decomposition is able to control the noise whereas the hysteresis threshold ensures proper connectivity. Computer generated and real images are tested for different fragment size in various noise background. The DNA sizing of fragments is compared with already existing method. The improved performances are analyzed using Figure of Merit, accuracy, sensitivity, specificity and false positive rate.