Feature-based classification of protein networks using confocal microscopy imaging and machine earning

Proc. Appl. Math. Mech.             online article

In this study machine learning was implemented in classification of CLSM images of FtsZ skeletons. First, images of two different FtsZ isoforms were used for morphological feature extraction. The features were used to train a classifier which was capable of classifying random, unlabeled FtsZ images with about 80% accuracy. The study shows that a combination of feature extraction algorithms and machine learning can enable efficient and quick classification of biological network- and signaling structures.