Method and device for automatic assessment of differences between normal and abnormal lungs

摘要

The invention concerns a method and device for automatic classification of the difference between normal and abnormal lung images, abnormal lung images being attributed to disease. Certain physical structural measurements are analysed, and a database for normal lungs is searched to determine whether these structural measurements are contained in it. The selected structural measurements are the oscillation in the root mean square R, and the first moment of the power spectrum M, for a lung structure. These two structural measurements are normalised using the database for normal lungs. A single structure index is determined from the two normalised structural measurements, by taking into account the distribution (or database) of the structural measurements obtained from abnormal lungs, to simplify the automatic classification of normal and abnormal lungs. Interesting areas are defined, and selected abnormal interesting areas are submitted to three different tests.