Classifying Fruit Fly Early Embryonic Developmental Stage Based on Embryo In situ Hybridization Images

In this paper, we present a supervised classification system for sorting Drosophila embryonic in situ hybridization (ISH) images according to their developmental stages. The proposed system first segments the embryo from an image and registers it for subsequent texture feature extraction. In order to extract the most distinguishing features for classifying developmental stages, we identify several areas of interest in an embryo with peculiar traits. Gabor filter is applied on these areas to extract texture features and Principal Component Analysis (PCA) is then performed on the extracted features to reduce dimensionality while retaining significant information. We adopt multi-class Support Vector Machine (SVM) as the classifier that learns model parameters from the training examples and classifies new examples with the trained model. We evaluate the system performance by comparing it to existing algorithms. The experimental results show that the proposed system achieves good performance in classifying Drosophila embryonic developmental stages and outperforms other state-of-the-art algorithms.