AUTOMATED RECOGNITION OF PATTERNS CHARACTERISTIC OF SUBCELLULAR STRUCTURES IN FLUORESCENCE MICROSCOPY IMAGES

Methods for numerical description and subsequent classification of cel lular protein localization patterns are described. Images representing the localization patterns of 4 proteins and DNA were obtained using f luorescence microscopy and divided into distinct training and test set s. The images were processed to remove out-of-focus and background flu orescence and 2 sets of numeric features were generated: Zernike momen ts and Haralick texture features, These feature sets were used as inpu ts to either a classification tree or a neural network. Classifier per formance (the average percent of each type of image correctly classifi ed) on previously unseen images ranged from 63% for a classification t ree using Zernike moments to 88% for a backpropagation neural network using a combination of features from the 2 feature sets, These results demonstrate the feasibility of applying pattern recognition methods t o subcellular localization patterns, enabling sets of previously unsee n images from a single class to be classified with an expected accurac y greater than 99%, This will provide not: only a new automated way to describe proteins, based on localization rather than sequence, but al so has potential application in the automation of microscope functions and in the field of gene discovery. (C) 1998 Wiley-Liss, Inc.