Tumor Spheres Quantification with Smoothed Euclidean Distance Transform.

ABSTRACT: Tumor sphere quantification plays an important role in cancer research and drugs screening. Even though the number and size of tumor spheres can be found manually, this process is time-consuming, prone to making errors, and may not be viable when the number of images is very large. This manuscript presents a method for automated quantification of spheres with a novel segmentation technique. The segmentation method relies on initial watershed algorithm which detects the minima of the distance transform and finds a tumor sphere for each minimum. Due to the irregular edges of tumor spheres, the distance transform matrix has often more number of minima than the true number of spheres. This leads to the over segmentation problem. The proposed approach uses the smoothed form of the distance tran

SUBMITTER: Sahin I 

PROVIDER: S-EPMC6179360 | biostudies-literature | 2018

REPOSITORIES: biostudies-literature