Semi-supervised segmentation using non-parametric snakes for 3D-CT applications in Radiation Oncology

We present a semi-supervised protocol for segmentation of tumors and normal anatomy for applications in Radiation Oncology. A primary goal in radiation therapy in oncology is to deliver high radiation dose to the perceived tumor while sparing the surrounding non-diseased organs. Consequently, a critical task in the workflow of radiation oncologists is the manual delineation of normal and diseased structures on 3D-CT scans. In this paper, we compare the results using a non-parametric snake technique with a gold standard consisting of manually delineated structures. Structures include tumors as well as normal organs including lungs, liver and kidneys. This technique provides fast segmentation that is robust with respect to noisy edges. In addition, this algorithm does not require the user to optimize a variety of parameters unlike many segmentation algorithms. We provide results that show the improvement in overlap between the manually delineated gold standard and the output of the segmentation algorithms using the user input.