Mammography is a screening tool for breast cancer detection that produces grayscale images of the breast. The fundamental problem in mammography image analysis is to extract the boundary of breast abnormality from its healthy background tissues. The process is also known as the image segmentation. The procedure is necessary for further clinical diagnosis and monitoring in Computer Aided Detection (CAD) analysis systems. Active contour method has been proven to be effective to extract boundary of an image. The recent and effective selective type of active contour model, termed Primal Dual Selective Segmentation (PDSS) model, was proposed in 2019. However, the PDSS model having problem in segmenting images with low contrast. It is known that low contrast image is commonly encountered in mammography images that can result to poor boundary extraction. Thus, the aim of this study is to modify the PDSS model to extract the boundary of abnormality region in mammography images. The modification is made by considering three different image enhancement algorithms which are histogram equalization, histogram stretching and adaptive histogram equalization as the new fitting terms in the PDSS model and these results in three variants of modified PDSS models termed as PDSS1, PDSS2 and PDSS3 respectively. The efficiency of the proposed models was then assessed by recording the computation time while the accuracy of the extracted image boundary was evaluated using the Jaccard (JSC) and Dice Similarity Coefficients (DSC). Numerical experiments demonstrated that the proposed PDSS2 model based on histogram stretching achieved the highest segmentation accuracy with the fastest computational speed compared to other models. In future, the proposed model can be extended into the threedimensional and colour formulations.