An automatic tool to identify bacterial types profiles using computer-vision and statistical modeling techniques is developed. Bacteriophage (phage)-typing methods are used to identify and extract representative profiles of bacterial types, such as the Staphylococcus Aureus. Currently, responses are read subjectively by a human expert. Large variability exists in the decision making process and the analysis is time-consuming, especially as technology is enabling the increase in the number of phages used for typing. The statistical methodology presented in this work, provides for an automated, objective and robust analysis of visual data, along with the ability to cope with increasing data volumes.
The input consists of scanned arrays of petri-dishes with the reactions of phages to the tested bacteria. The data is automatically segmented utilizing statistical modeling via Gaussian mixture models and Expectation-Maximization (EM) learning. The segmentation process is used to separate signal regions from the background, and to automatically align the reactions to a grid for reaction localization. Gaussian mixture modeling is used in a selected feature space for the probabilistic categorization of each local region into positive and negative reactions. A final phage profile is generated per group of image inputs based on a statistical analysis of the group reactions. The statistical methodology presented in this work may be applicable in additional related domains, such as microarray data.
For an overview of the method see the following Slides.
For problems or
questions regarding this web page contact