A Framework for Object Detection and Segmentation in Medical Images Using Arbitrary Shapes

Detection and segmentation objects are crucial to extract useful information from medical images [1] and still challenging tasks in CAD software development. Due to the irregular structure and high variability of objects of interest, there is no universal solution for detecting objects and segmenting them in all kinds of medical images. This work is a part of a larger project aimed mostly at detection, coarse segmentation and visualization of given objects in microscope images. The purpose of this paper is to introduce a robust framework to facilitate detecting objects with arbitrary size and shape and segmenting them in different kinds of medical images with the help of a library of irregular smooth shapes.