An adaptive threshold method for mass detection in mammographic images

An early detection of abnormalities is the key point to improve the prognostic of breast Cancer. Masses are among the most frequent abnormalities. Their detection is however a very tedious and time-consuming task. This paper presents an automatic scheme to perform both detection and segmentation of breast masses. Firstly, the breast region is determined and extracted from the whole mammogram image. Secondly, an adaptive algorithm is proposed to perform an accurate identification of the mass region. Finally, a false positive reduction method is applied through a feature extraction method and classification using the advantages of multiresolution representations (curvelet and wavelet). The classification step is achieved using SVM and KNN classifiers to distinguish between normal and abnormal tissues. The proposed method is tested on 118 images from mammographic images analysis society (MIAS) datasets. The experimental results demonstrate that the proposed scheme achieves 100% sensitivity with average of 1.87 False Positive (FP) detections per image.