Classification of Medical Images Using the Model for Formation of Viewing Trajectory

摘要

This work considers how the research vision behavior model is used to analyze medical images. An viewing trajectory was formed for each image; the distribution of the frequencies of its angular parameters and the distribution of the considered fields were estimated. The correlation coefficient for images of the same class varied within 0.62–0.89, whereas the correlation coefficient in comparing images of different classes was less than 0.3.