< Back to Article

Ultra High Content Image Analysis and Phenotype Profiling of 3D Cultured Micro-Tissues

Figure 4

Characterization of cellular phenotype by clustering and classification.

(a) Hierarchical clustering result using an average matrix as distance matrix. The scale of dendrogram is the natural logarithm of . (b) Five defined classes of test compounds and corresponding compounds and number of data points. (c) Classification result using multiple classification methods. Feature selection with search algorithm “forward” and criterion “Mahalanobis distance” was applied to detect optimal number of features. For each classification method and each number of selected features, 10 fold cross-validation was repeated 10 times, resulting in 10 error rates. The average error rates are shown in the chart with standard deviation as error bar. SVC means support vector machine classification.

doi: https://doi.org/10.1371/journal.pone.0109688.g004