A statistical measure for effects of treatment for MTB derived from comparison of digital x-ray images

A series of chest radiograph images will be compared for purposes of monitoring effect of treatment on a given Mycobacterium Tuberculosis (MTB) patient. A subset of the first image corresponding to the patients first visit to the clinic will be selected as being representative of the infected areas. The corresponding subset from the second image (second visit) will be similarly obtained, and then intensity image registration between these two subsets is carried out, using correlations. This procedure is repeated for all available images. For each subset-image the grey level histogram is obtained and compared. This study shows that the coefficient of determination for the Unreplicated Linear Functional Relationship model, RF² are superior to the ordinary correlation RS² for image registration. Finally, a comparison of two histogram is achieved by comparing appropriate measures of location and spread.