Mammographic breast density classification using a deep neural network: Assessment based on inter-observer variability

摘要

Mammographic breast density is an important risk marker in breast cancer screening. The ACR BI-RADSguidelines (5th ed.) dene four breast density categories that can be dichotomized by the two super-classesdense" and ot dense". Due to the qualitative description of the categories, density assessment by radiologistsis characterized by a high inter-observer variability.To quantify this variability, we compute the overall percentage agreement (OPA) and Cohen's kappa of 32radiologists to the panel majority vote based on the two super-classes. Further, we analyze the OPA betweenindividual radiologists and compare the performances to an automated assessment via a convolutional neuralnetwork (CNN). The data used for evaluation contains 600 breast cancer screening examinations with four viewseach. The CNN was designed to take all views of an examination as input and trained on a dataset with 7186cases to output one of the two super-classes.The highest agreement to the panel majority vote (PMV) achieved by a single radiologist is 99%, the lowestscore is 71% with a mean of 89%. The OPA of two individual radiologists ranges from a maximum of 97.5% toa minimum of 50.5% with a mean of 83%. Cohen's kappa values of radiologists to the PMV range from 0.97 to0.47 with a mean of 0.77. The presented algorithm reaches an OPA to all 32 radiologists of 88% and a kappa of0.75.Our results show that inter-observer variability for breast density assessment is high even if the problem isreduced to two categories and that our convolutional neural network can provide labelling comparable to anaverage radiologist. We also discuss how to deal with automated classication methods for subjective tasks.