Automated recognition system to classify subcellular protein localizations in images of different cell lines acquired by different imaging systems.

Systemic analysis of subcellular protein localization (location proteomics) provides clues for understanding gene functions and physiological condition of the cells. However, recognition of cell images of subcellular structures highly depends on experience and becomes the rate-limiting step when classifying subcellular protein localization. Several research groups have extracted specific numerical features for the recognition of subcellular protein localization, but these recognition systems are restricted to images of single particular cell line acquired by one specific imaging system and not applied to recognize a range of cell image sources. In this study, we establish a single system for automated subcellular structure recognition to identify cell images from various sources. Two different sources of cell images, 317 Vero (http://gfp-cdna.embl.de) and 875 CHO cell images of subcellular structures, were used to train and test the system. When the system was trained by a single source of images, the recognition rate is high and specific to the trained source. The system trained by the CHO cell images gave high average recognition accuracy for CHO cells of 96%, but this was reduced to 46% with Vero images. When we trained the system using a mixture of CHO and Vero cell images, an average accuracy of recognition reached 86.6% for both CHO and Vero cell images. The system can reject images with low confidence and identify the cell images correctly recognized to avoid manual reconfirmation. In summary, we have established a single system that can recognize subcellular protein localizations from two different sources for location-proteomic studies. studies.

Boosting multiclass learning with repeating codes and weak detectors for protein subcellular localization.

MOTIVATION: Determining locations of protein expression is essential to understand protein function. Advances in green fluorescence protein (GFP) fusion proteins and automated fluorescence microscopy allow for rapid acquisition of large collections of protein localization images. Recognition of these cell images requires an automated image analysis system. Approaches taken by previous work concentrated on designing a set of optimal features and then applying standard machine-learning algorithms. In fact, trends of recent advances in machine learning and computer vision can be applied to improve the performance. One trend is the advances in multiclass learning with error-correcting output codes (ECOC). Another trend is the use of a large number of weak detectors with boosting for detecting objects in images of real-world scenes. RESULTS: We take advantage of these advances to propose a new learning algorithm, AdaBoost.ERC, coupled with weak and strong detectors, to improve the performance of automatic recognition of protein subcellular locations in cell images. We prepared two image data sets of CHO and Vero cells and downloaded a HeLa cell image data set in the public domain to evaluate our new method. We show that AdaBoost.ERC outperforms other AdaBoost extensions. We demonstrate the benefit of weak detectors by showing significant performance improvements over classifiers using only strong detectors. We also empirically test our method's capability of generalizing to heterogeneous image collections. Compared with previous work, our method performs reasonably well for the HeLa cell images. AVAILABILITY: CHO and Vero cell images, their corresponding feature sets (SSLF and WSLF), our new learning algorithm, AdaBoost.ERC, and Supplementary Material are available at http://aiia.iis.sinica.edu.tw/