Segmentation of tumor objects in H&E images using graph-based clustering based on opponent color space and internal image statistics Author: Luong Nguyen Email: lun5@pitt.edu Started 10/23/2014
Main components:
- Inputs: tiling of WSI, parallelization, rotation matrix
- Annotations: collect annotations from pathologists
- Affinity calculation: Features and affinity function
- Clustering: NMinCut, Embedded Spectral Clustering
Reused codes and packages from Citation of KDE Citation of Multiscale Spectral Clustering Citation of Crisp Boundary