Evaluation of Segmentation Algorithms on Cell Populations Using CDF Curves

Published

February 24, 2012

Author(s)

Robert C. Hagwood, Javier Bernal, Michael W. Halter, John T. Elliott

Abstract

Cell segmentation is a critical step in the analysis pipeline for most imaging cytometry experiments and the segmentation algorithm can effect the quantitative data derived from image analysis. Methods to evaluate segmentation algorithms are important for aiding the selection of segmentation algorithms. We describe the application of misclassification error rates in evaluating the performance of four common segmentation algorithms. Misclassification error is an important criterion for the analysis of quantitative descriptors of cell morphology involving pixel counts, e.g. projected area, aspect ratio, diameter which are highly sensitive to the accuracy of the segmentation scheme. Since the cumulative distribution function (CDF) captures completely the stochastic properties of a population of misclassification errors it is used to compare segmentation performance.

Citation

Computer Vision and Image Understanding

Volume

31

Issue

2

Pub Type

Journals

Download Paper

Local Download

Keywords

segmentation, fluorescence microscopy, image cytometry, cell morphology, misclassification errors, flow cytometry, discriminant analysis.

Citation

Hagwood, R. , Bernal, J. , Halter, M. and Elliott, J. (2012), Evaluation of Segmentation Algorithms on Cell Populations Using CDF Curves, Computer Vision and Image Understanding, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=907743 (Accessed November 21, 2024)