Confidence Estimation in Stem Cell Classification

Published

October 15, 2015

Author(s)

Peter Bajcsy, Jana Kosecka, Zahra Rajabi

Abstract

We study the problem of supervised classification of stem cell colonies and confidence estimation of the attained classification labels. The problem is investigated in the application context of heterogeneity labels of stem cell colonies observed by using fluorescent microscopy imaging. Given the colony images and their characterization using numerous image statistics, we report the classification results using adaptive k-Nearest Neighbor (NN) algorithm. This algorithm minimizes typical k-NN classification bias by giving more weight to more informative features in predicting class posterior probabilities. We then estimate the confidence of each prediction for unlabeled data using p-value and strangeness metrics. We show that such an introspective ability can gradually increase the learned model accuracy, quantify false positives, and guide the resource-limited manual colony annotation process to provide training labels for the less confident unlabeled samples.

Proceedings Title

BioImage Informatics Conference 2015

Conference Dates

October 14-16, 2015

Conference Location

Gaithersburg, MD, US

Pub Type

Conferences

Download Paper

Local Download

Keywords

classifiers, uncertainty, bias

Citation

Bajcsy, P. , Kosecka, J. and Rajabi, Z. (2015), Confidence Estimation in Stem Cell Classification, BioImage Informatics Conference 2015, Gaithersburg, MD, US, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=918718 (Accessed October 13, 2025)