Histogram Analysis

The shape of the histogram can be represented by the moments of the distribution about the mean gray level. Among the histogram-based features that have been found to be useful is the third moment [25,26], which characterizes the asymmetry or skewness of the histogram. To be more sensitive to local variations, the calculation is performed for regions of the image based on local means [27].

To generate the histogram, a method similar to that described in the discussion of PD can be used to identify the edge of the breast and areas to be excluded, so that only pixels lying between the chest wall and the skin edge (i.e., that are within the breast) are counted and assigned to h. For an image

FIGURE 4 Screen presentation of a program developed for interactive analysis of mammographie density. See also Plate 49.

(or region of an image) containing N pixels, the fcth moment, mk, of the histogram is given by mk

0 0

Post a comment