Robust Edge-Stop Functions for Edge-Based Active Contour Models in Medical Image Segmentation

Medical Image Segmentation with Adjustable Computational Complexity Using Data Density Functionals

“…To the purpose of medical image segmentation, the edge-based active contour models [5,9,10,48] and the machine learning methods [5,9,32,[48][49][50][51][52][53] are two main solutions regarding the medical image segmentation. The edge-based active contour models were developed based on the concept of energy minimization, starting with the well-known snake model [54,55].…”

“…Then, a level set equation is used to estimate the evolution of the level set function. The level set equation often includes an edge stop function with a Gaussian kernel, a potential function, and several energy parameters regarding the distance regularization energy, the length terms, and the area term [48]. Additionally, the energy parameters always need to be estimated by experiments or simulations.…”

“…Meanwhile, the users have to initially define the foreground and background to train the classifiers by themselves. The level set function typically requires re-initialization to avoid irregularities during its evolution [48]. Thus, the naïve active contour models have inevitable inconvenience on the issues of determination of the energy parameters, the intrinsic algorithmic properties, and the intervention of marking the foreground and background of the region of interest.…”

“…To circumvent these deficiencies, state-of-the-art machine learning methods in probabilistic perspectives have brought several useful techniques [5,9,10,32,[48][49][50][51][52][53]. By combining the graph theory with topological priors [53], morphologies of three-dimensional biological images could be delineated and reconstructed visually.…”

Eyebrow Modeling For Knowlege Representation

“…Currently, the pure eyebrow images are produced generally by manually cropping [1], [3]- [6] or automatic segmentation [7]. Broadly speaking, the level set method (LSM) has been applied extensively in extracting objects due to its abilities to account for topological variations and convergence stability [8]- [12], [14], [16]. In process of implementing LSM, the initial-curve is a key factor for level set evolution, and it highly depends on the appropriate manual initialization [12].…”