Standardizing breast density assessments

Abstract

A method, system and computer program product for determining changes in breast density. A generative adversarial network is trained to predict an appearance of a mammogram image for a patient's current examination based on mammogram images assigned labels of a first type of density classification. An appearance of a mammogram image for a patient's current examination is predicted using the generative adversarial network based on a mammogram image(s) obtained from the patient's prior examination assigned labels of the first type of density classification. A comparison is made between the predicted and actual mammogram images for the patient's current examination to determine if there is a difference between scores assigned to the predicted and actual mammogram images, and if so, whether such difference can be attributed to the subjective assessment by different physicians or changes in the standards of density classification or is due to a real change in breast density.

1 . A system, comprising:
 a memory for storing a computer program for determining changes in breast density in mammogram images; and   a processor connected to said memory, wherein said processor is configured to execute program instructions of the computer program comprising:
 receiving a set of mammogram images labeled by users with a second type of breast density classification, wherein said mammogram images comprise mammogram images of patients obtained from both prior and current examinations; 
 assigning labels of a first type of breast density classification to said received set of mammogram images using a convolutional neural network trained to predict labels of said first type of breast density classification in mammogram images; 
 training a generative adversarial network to predict an appearance of a mammogram image for a patient's current examination based on said received set of mammogram images assigned labels of said first type of breast density classification; and 
 predicting an appearance of a mammogram image for said patient's current examination using said generative adversarial network based on a mammogram image obtained from a prior examination of said patient labeled with said first type of breast density classification. 
   
     
     
         2 . The system as recited in  claim 1 , wherein the program instructions of the computer program further comprise:
 predicting said appearance of said mammogram image for said patient's current examination using said generative adversarial network based on said mammogram image obtained from said patient's prior examination and a feature vector representing demographic characteristics of said patient.   
     
     
         3 . The system as recited in  claim 1 , wherein the program instructions of the computer program further comprise:
 comparing said predicted appearance of said mammogram image for said patient's current examination with an actual mammogram image for said patient's current examination labeled with said first type of breast density classification.   
     
     
         4 . The system as recited in  claim 3 , wherein said predicted mammogram image is assigned a score for said first type of breast density classification, wherein said actual mammogram image is assigned a score for said first type of breast density classification. 
     
     
         5 . The system as recited in  claim 4 , wherein the program instructions of the computer program further comprise:
 accepting labels of said first type of breast density classification in mammogram images of said patient for past and current examinations in response to said score assigned to said predicted mammogram image matching said score assigned to said actual mammogram image.   
     
     
         6 . The system as recited in  claim 4 , wherein the program instructions of the computer program further comprise:
 generating an indication indicating that a difference in value between said score assigned to said predicted mammogram image and said score assigned to said actual mammogram image is due to subjective assessments by different physicians or changes in standards of density classification in response to said score assigned to said predicted mammogram image differing from said score assigned to said actual mammogram image within a threshold amount of difference.   
     
     
         7 . The system as recited in  claim 4 , wherein the program instructions of the computer program further comprise:
 generating an indication requesting a technician to analyze labels for mammogram images for said patient's prior and current examinations labeled with said first type of breast density classification to determine if a difference in value between scores assigned to said predicted mammogram image and said actual mammogram image is due to subjective assessments by different physicians or changes in standards of density classification or is due to a real change in breast density in response to said score assigned to said predicted mammogram image differing from said score assigned to said actual mammogram image not within a threshold amount of difference.   
     
     
         8 . The system as recited in  claim 1 , wherein said first and second types of breast density classifications comprise different standards of a Breast Imaging Reporting and Data System (BI-RADS) breast density classification. 
     
     
         9 . The system as recited in  claim 1 , wherein said first type of breast density classification comprises BI-RADS fifth edition density labels, wherein said second type of breast density classification comprises BI-RADS fourth edition density labels. 
     
     
         10 . A computer-implemented method for standardizing breast density classifications in mammogram images, the method comprising:
 receiving a set of mammogram images labeled by users with a first type of breast density classification;   training a convolutional neural network to predict labels of said first type of breast density classification in mammogram images using said received set of mammogram images; and   assigning labels of said first type of breast density classification to mammogram images labeled under a second type of breast density classification using said convolutional neural network.   
     
     
         11 . The method as recited in  claim 10 , wherein said first and second types of breast density classifications comprise different standards of a Breast Imaging Reporting and Data System (BI-RADS) breast density classification. 
     
     
         12 . A computer program product for standardizing breast density classifications in mammogram images, the computer program product comprising one or more computer readable storage mediums having program code embodied therewith, the program code comprising programming instructions for:
 receiving a set of mammogram images labeled by users with a first type of breast density classification;   training a convolutional neural network to predict labels of said first type of breast density classification in mammogram images using said received set of mammogram images; and   assigning labels of said first type of breast density classification to mammogram images labeled under a second type of breast density classification using said convolutional neural network.   
     
     
         13 . The computer program product as recited in  claim 12 , wherein said first and second types of breast density classifications comprise different standards of a Breast Imaging Reporting and Data System (BI-RADS) breast density classification.