Automated mammographic density estimation and display method using prior probability information, system for the same, and media storing computer program for the same

An automated mammographic density estimation method using prior probability information, the method comprising steps of:

• constructing a population-based tissue probability map (PTPM) by extracting tissue probability information from a plurality of image area previously segmented by at least one expert according to the dense and fatty regions in a preprocessed mammogram of a population;
• and
• estimating the density by applying the population-based tissue probability map (PTPM) to a preprocessed interest mammogram to obtain pixel-wise tissue probability followed by calculating the percent mammographic density,
• wherein extracting the tissue probability information in the constructing the population-based tissue probability map (PTPM) comprises calculating local statistical feature set from the plurality of region of interest sampled within the image area previously segmented by the expert according to the dense and fatty tissue types, and obtaining tissue probability density functions for the dense and fatty tissue types in a local statistical feature space,
• wherein the tissue probability density function is obtained by a nonparametric probability density estimation scheme, and
• wherein the nonparametric probability density estimation scheme comprises a Parzen Window method.

A method of displaying the mammographic density estimation results comprising:

• generating an evidential image in which the basis data of respective pixels in the intermediate estimation result are color coded and overlaid on the interest mammogram;
• and
• displaying text characters including to the numerical values of estimation results on the evidential image,
• wherein the mammographic density estimation results are obtained with an automated mammographic density estimation method using prior probability information, the automated mammographic density estimation method comprising steps of:
• constructing a population-based tissue probability map (PTPM) by extracting tissue probability information from a plurality of image area previously segmented by at least one expert according to the dense and fatty regions in a preprocessed mammogram of a population;
• and
• estimating the density by applying the population-based tissue probability map (PTPM) to a preprocessed interest mammogram to obtain pixel-wise tissue probability followed by calculating the percent mammographic density,
• wherein extracting the tissue probability information in the constructing the population-based tissue probability map (PTPM) comprises calculating local statistical feature set from the plurality of region of interest sampled within the image area previously segmented by the expert according to the dense and fatty tissue types, and obtaining tissue probability density functions for the dense and fatty tissue types in a local statistical feature space, and
• wherein constructing the tissue probability map from a tissue probability density function in the constructing the population-based tissue probability map (PTPM) is derived by a proportion of each tissue probability density function to the sum of the dense and fatty tissue probability density functions.

An automated mammographic density estimation system for carrying out an automated mammographic density estimation method comprising:

• constructing a population-based tissue probability map (PTPM) by extracting tissue probability information from a plurality of image area previously segmented by at least one expert according to the dense and fatty regions in a preprocessed mammogram of a population;
• and
• estimating the density by applying the population-based tissue probability map (PTPM) to a preprocessed interest mammogram to obtain pixel-wise tissue probability followed by calculating the percent mammographic density,
• wherein extracting the tissue probability information in the constructing the population-based tissue probability map (PTPM) comprises calculating local statistical feature set from the plurality of region of interest sampled within the image area previously segmented by the expert according to the dense and fatty tissue types, and obtaining tissue probability density functions for the dense and fatty tissue types in a local statistical feature space, and
• wherein constructing the tissue probability map from a tissue probability density function in the constructing the population-based tissue probability map (PTPM) is derived by a proportion of each tissue probability density function to the sum of the dense and fatty tissue probability density functions.

A non-transitory media for storing a computer program, wherein the computer program stored in the media comprises a computer algorithm provided for implementing an automated mammographic density estimation method comprising:

• constructing a population-based tissue probability map (PTPM) by extracting tissue probability information from a plurality of image area previously segmented by at least one expert according to the dense and fatty regions in a preprocessed mammogram of a population;
• and
• estimating the density by applying the population-based tissue probability map (PTPM) to a preprocessed interest mammogram to obtain pixel-wise tissue probability followed by calculating the percent mammographic density,
• wherein extracting the tissue probability information in the constructing the population-based tissue probability map (PTPM) comprises calculating local statistical feature set from the plurality of region of interest sampled within the image area previously segmented by the expert according to the dense and fatty tissue types, and obtaining tissue probability density functions for the dense and fatty tissue types in a local statistical feature space, and
• wherein constructing the tissue probability map from a tissue probability density function in the constructing the population-based tissue probability map (PTPM) is derived by a proportion of each tissue probability density function to the sum of the dense and fatty tissue probability density functions.