Wavelet-based quantitative evaluation of a digital density equalization technique in mammography

Abstract

In this study, quantitative evaluation of a proposed digital density equalization technique in mammography was carried out. The evaluation was performed on a set of 90 mammograms, based on wavelet-generated measurable parameters of image quality, such as contrast, noise and contrastto- noise ratio (CNR). These parameters were estimated for dense mammary gland and breast periphery, for both initial and corresponding corrected mammograms. The equalization character of the technique was also examined. A statistically significant (p<0.05) or highly significant increase (p<0.0005) was observed in breast periphery and mammary gland contrast, noise and CNR values. The proposed technique was found to result in density equalization, since the decrease in equalization index is statistically highly significant (p<0.0005).