Image-processing based taxonomy analysis of bacterial macromorphology using machine-learning models

Classification of bacteria is essential in the medical diagnosis of infectious agents, their phylogenetic study, and their biotechnological exploitation for healthcare, food, industry, and agricultural sectors. Nevertheless, skilled experts and professional human-effort are necessary to identify and classify the bacteria manually. With the advancement of technology, now the task of recognizing details from digital stereomicroscopes is being performed by computers based on machine learning and computer vision technologies. Besides, machine learning methods include Deep Neural Networks (NN) has attained remarkable outcomes in the field of image classification recently. Furthermore, other machine learning methods except for NN methods already have acceptable performance. In this paper, we review the publications, which investigate the discrimination between bacteria genera and suborders based on macroscopic images via image processing and machine learning methods. The published research works in this regard are summarized, and the pros and cons of them are discussed. Moreover, the related databases and resources for this purpose are surveyed, and the lack of such data points in the global catalogue of microorganisms is criticized. In addition, in this paper, we have investigated an approach to automate the process of bacteria recognition and classification with the use of the Gabor transform and XGBoost classification method. We have used a dataset that includes microscopic images of three different Myxobacterial suborders. The trained model was able to recognize and classify all members of three different categories of bacteria, while the experimental results of prediction achieved an accuracy of around 91%, which has enhancement about 2% in term of accuracy.