Web-Based Multi-Observer Segmentation Evaluation Tool

Multi-observer segmentation evaluation is useful in the imaging community. We have developed a web-based software application for automatic performance evaluation of multiple image segmentations, based on the Baysian Decision framework. It computes a probabilistic estimate of the true segmentation (ground truth map) and performance measures for the individual segmentations (sensitivity and specificity). The strength of the tool is that it integrates two kinds of prior knowledge about segmentations: the truth prior (the prior probability of different segmentations) and the observer prior (the performance measures of observers), which can generate more accurate evaluations.

• Type: Conference
• Year: 2008
• Where: CBMS
• Authors: Yaoyao Zhu, Xiaolei Huang, Daniel P. Lopresti, L. Rodney Long, Sameer Antani, Zhiyun Xue, George R. Thoma
• Keywords: CBMS 2008, Medical Imaging, Multi-observer Segmentation Evaluation, Multiple Image Segmentations, Performance Measures