Automatic segmentation of mammographic density

Rationale and Objectives. The purpose of this study was to evaluate a compl etely automatic method, based on Kittler's optimal threshold, to estimate b reast density by using the mammographers' definition. Materials and Methods. Thirty-two normal, right-craniocaudal-view mammogram s of women aged 37-86 years were digitized. The whole breast area was segme nted by using Kittler's optimal threshold procedure, and the dense portions were then segmented by using a modified version of Kittler's method. Segme ntation results were validated by three independent mammographers who provi ded a signed percentage tin steps of 5%) to indicate the difference between their own visual estimation of the dense portions and the results obtained with the algorithm. The difference between the algorithm measurements and the mammographers' measurements was compared to the interobserver differenc es. Results. A high correlation was found between the algorithm measured densit y and the mammographers' measurements. Spearman correlations ranged from 0. 92 to 0.95 (P < .001). Algorithm-measured density differed from the mammogr aphers' measurements by an average of 6.9% tie. average of the absolute dif ferences). In contrast, mammographers' measurements differed between themse lves by an average of 5.4%. Conclusion. The difference between density as measured with the algorithm a nd as measured by the mammographers is similar to the differences observed between mammographers. This algorithm could be useful in providing clinical ly accurate estimates of breast density.