An integrated DWT-wiener and marker controlled watershed driven active contours for nuclear segmentation in Histopathological Images

Nuclear segmentation is one of the challenging issues in the field of Medical Image segmentation for Histopathological Images. Various edge based and region based approaches have been proposed in the literature. Both the approaches suffer from deficiencies such as poor edge information and over/under segmentation respectively. In the recent past, Active contours have emerged as a powerful techniques for effective image segmentation. Many of the techniques proposed have posed the problem of contour initialization, computation of gradient information and extraction of overlap resolutions as the major challenge. To address these issues, a novel strategy of driving the contour evolution based on integrated discrete wavelet transform based wiener filter and marker controlled watershed approach has been proposed. The experimentation was conducted on few BC tissue images and the efficacy of the approach has been studied based on measures of object detection accuracy and overlap resolution in comparison with the ground truth obtained from the pathologist in the Department of Pathology, JSS Hospital.