Biological specimens such as the cornea, retina, skin etc., consist of multiple layers, each layer having certain characteristic properties. The state-of-the-art imaging modalities extract the multilayer information using different techniques that rely on the optical properties of the layers. The tomograms obtained can be quantified by parameters such as the layer thickness, region areas, moments and interlayer spacing. The objective of this project is to develop efficient active-contour modeling (also known as snakes) algorithms for estimating these parameters. In particular, we are interested in parametric active contours that are parsimoniously represented using B-spline basis functions. Innovation lies in the specification of a suitable multilayer model in the framework of splines, and definitions of appropriate cost functions which when optimized enable the contours to lock onto the structure of interest.
Pre-requisite: Knowledge of Image Processing and good programming skills. The software will have to be developed in the form of a plugin for ImageJ, taking advantage of existing fully functional optimization engine and graphical user interface developed in our laboratory.