Tumor Boundary Delineation Using Abnormality Outlining Box Guided Modified GVF Snake Model

Abstract

Segmentation is a million dollar task in order to highlight any region of interest (ROI) such as tumor, bleed, edema, or infarct in any medical image. Disease diagnosis, prognosis, surgery, and rehabilitation aspects are guided by segmentation results in oncology. The best ideology is to use high-end active contour models for tumor detection, location, and delineation. Traditional parametric active contour model like GVF (gradient vector flow) snake model suffers from more sensitive parameters, limited capture range, more sensitive to noise, and indentation while segmenting brain tumors in medical images. In order to address and solve these problems, a modified GVF is proposed. The modified snake model proposes a new design for the snake external force to resolve these problems by inducing a strong force near edges. This paper will address how an abnormality outlining box (AOB) guides modified GVF to get rid of demerit like lack of choice of precise initial contour. AOB-guided modified GVF is an integration of AOB and modified GVF external force. The experimental results prove the modified snake model captures the tumor efficiently and quickly even in the presence of any noise artifacts.