An automated detection method for partial loss masses existing around mammary-gland region on mammograms

摘要

We have been developing an automated detection algorithm to detect the partial loss masses around the thick-mammary gland region on mammograms. In this study, we have improved the performance of the previous method, using the analytical result of the distribution of the fibroglandular breast tissue density, and changing 4 sector-shaped mass patterns to 8 sector-shaped mass patterns. To compare the new method with the previous one, we examined the FROC curves for both methods on 161 mammograms. The detection sensitivity was changed from 62% to 90% at keeping the 0.95 false positive per image.