Semantic Image Processing for Developmental Data Science.

ImageJ is an OpenSource image processing tool written in Java which has been extremely helpful in the analysis of scientific images especially medical and microscopic. ImageJ package consists of a plugin for segmenting all the cells in an embryo of a C.Elegan from raw SPIM images. This plugin which has been built as a part of GSoC’17 comprises of standard image processing techniques to segment the cells.

While attempting to segment cells from High-Resolution time-lapse movies of Embryogenesis, the used unsupervised algorithms like Intensity thresholding, Watershed Transform, Active Contours, etc, which depend on extracted local or global features, fail to be resistant against uneven illumination, optical noises, complex cellular shapes and more of such distractions. Thus, we are still in the pursuit of having a robust method for accurately segmenting microscopy images.

The top priorities of this proposal are: Extend the plugin’s capability to more accurately segment the cells by adding Semi-Supervised and Unsupervised methods. Focus on extending capabilities of tools for developmental DataScience with the help of datasets or develop a cell-tracking system for Bright-field movies.