Medical Image Synthesis via Monte Carlo Simulation

A large number of test images and their "ground truth" segmentations are needed for performance characterization of the many image segmentation methods. In this work we developed a methodology to form a probability distribution of the diffeomorphism between a segmented template image and those from a population, and consequently we sample from these probability distributions to produce test images. This method will be illustrated by producing simulated 3D CT images of the abdomen for testing the segmentation of the human right kidney.

• Type: Conference
• Year: 2002
• Where: MICCAI
• Authors: James Z. Chen, Stephen M. Pizer, Edward L. Chaney, Sarang C. Joshi
• Keywords: Image Segmentation Methods, Medical Imaging, MICCAI 2002, Segmented Template Image, Test Images