Interactive edge detection and tracing in cellular nuclei images using B-splines and neural network

During the analysis of histological images of cancer tissue, the specialist must carefully look for cellular shapes and sizes, as well as orientation, in order to make a diagnosis. This work develops an interactive auxiliary method to help the physician to quickly trace the edge for each cell, as selected by him (her) in a microscopic image. The method is based in a gradient and gray-level study of the image, along a series of radials centered in a selected cell. The analysis outputs a set of points from which the edge points are selected. The edge tracing is then performed using the B-spline method.