The Chan-Vese segmentation algorithm is robust and has been used to segment different types of images. This algorithm relies on the global properties of an image (gray level intensities in regions, length of contours, area of ​​regions), so it is more suitable in cases where contour information is not very predominant . The results are qualitatively good for noisy images and images with complicated topologies. By reviewing the state of the art of segmentation techniques, we find that this method is widely used in segmenting regions in medical images. Low contrast images do not have well-defined homogeneous regions, performance may be degraded in these scenarios since the method assumes homogeneous foreground and background intensities. The level sets being non-parametric, include a regularization term such as penalty on curve/length/area or curvature. These regularization terms do not contain any information about the shape of the region of interest. The Chan-Vese algorithm can be slow for some applications. Depending on the type and size of the image and the number of iterations required, segmentation can be slow, hence GPU implementations of the algorithm reduce the segmentation problem to an ordinary differential equation rather than 'to a nonlinear PDE [13] can be used to speed up the algorithm. in 3D. Currently, the contour segment parts of the model are manually selected and grouped into groups. It is important to automate the part batch learning step. The temporal performance of the algorithm is not discussed in the paper.4. FUTURE WORK4.1. Active contours without edges4.1.1. Selecting parameters and initializing the Level Set functionThis method is semi-automatic. This requires user interaction to adjust parameters and initialize the level set. To obtain good quality segmentations, the parameters must be adjusted for each image. Machine learning algorithms can be used to learn parameter values ​​for homogeneous images from a set of training images. Unfortunately, actual images are not consistent, there may be variations in the image due to different lighting conditions. Methods designed to automatically adjust these parameters would benefit the segmentation process. Careful initialization of the level set is necessary, as it has a direct correlation with the time required for convergence. Atlas-based initializations can also be used to segment structures from medical images.4.1.2. Improve efficiencyAccuracy and precision can be improved by incorporating prior information about the target to be segmented. To increase computing efficiency, multi-scale processing and parallelizations are viable solutions.3.2. Shape-guided edge clustering with particle filters The edge clustering algorithm was applied to the ETHZ shape classes. ETHZ shape classes exhibit significant intra-class variations, scale changes, and illumination changes. The results were shown in the presence of a lot of disorder in the