The value of immage autosegmentation with modofied region growing algorithm in disgnosis of breast masses
Wang Jiawan · Jiangsu Medical Journal · 2011
Objective To investigate the value of image autosegmentation with modified region growing algorithm(MRGA) in diognosis of breast masses.Methods The images of 172 pieces of breast masses were segmentated with MRGA and manual manner at the same time and the overlap ratio(η) and mean distance(Md) from the segmentation region were compared between two methods.Taking the resuls from manual segmentation as a golden standard,the sensitivity and accuracy of MRGA were determined.The spiculation index(SI) of benign and malignant mass image was calculated and compared.Results Of 172 masses images,159 masses(92.4%) were automatically detected with MRGA.The overlap ratio of mass regions showed by two algorithms was 83.7 % and the mean distance was(0.86±0.03) mm.The SI value of 62 benign masses was 45.07±11.55,which was lower than 67.69±9.19 of 97 malignant masses(P0.01).Conclusion Breast mass region can be extracted with MRGA and speculation index can be used to discriminate preliminarily benign breast mass from the malignant,which benefits early diagnosis of breast cancer.