[en] The term unevenness or inhomogeneity has been used rather vaguely in nuclear medicine to characterize the radioactive distribution in organ images. The purpose of this study was to establish methodology to numerize the unevenness on images and to find the physiological and functional aspects included in uneven images. Using the count profiles of each of the rows and/or columns derived from image matrix data, many indexes, such as size, deposition characteristics and indexes of unevenness of count distribution were defined and computed. The values derived by analyzing count profiles of each row were named horizontal variables. Similar procedures were performed vertically for each column and the values thus obtained were named vertical variables. To evaluate the unevenness of images, two different approaches were tried. One was H + V analysis: the indexes were taken from those size and deposition characteristics, horizontal and vertical variables and their respective means, standard deviation, coefficient of variation and lengths. Thus the total number of indexes was 33 in H + V analysis. Another approach was HV analysis: here disregarding the rows and columns, similar calculations were made for the combined rows and columns for the H and V variables. The total number of indexes was 21 in HV analysis. We applied this method to lung perfusion images and found that at least more than 250 counts per cell (pixel) were necessary for the analysis of unevenness. (author)