Acquisition of mammogram breast border reference data from non-experts
Ramachandran Chandrasekhar, Yianni Attikiouzel · 2003
Automatic segmentation needs validation by expert radiologists, who are few and often very busy. This paper presents an approach to manually acquiring breast border reference data from non-experts using freely available image display and manipulation programs like xv and Gimp. The resulting reference images may then be compared with those from automatic segmentation and the degree of mismatch may be used to quantify the quality of segmentation. Preliminary results show the feasibility of this approach.