A partial supervised mammographic masses detection method
Dong Yan · Zhongguo yixue yingxiang jishu · 2005
Objective Accurate selection of suspected masses in medical images. Methods A novel method inspired by active relevance feedback is introduced as partial supervised mass detection. On the basis of pre-detection, partial suspected regions provided by active feedback are labeled by the expert (or the users) as the positive or negative sample regions. After a few training and learning interactively, the final mass detection result is given. Results The method can accurately select the suspected masses through eliminate the false-positive regions. Conclusion The method has good accuracy and helps to further analysis of the computer-aided diagnosis system.