Abstract: A novel approach to computer aided microcalcifications detection in mammograms was presented. Firstly dome extraction was accomplished by using morphological grayscale reconstruction, which could not only remove the breast structures but also remain microcalcifications commendably. Secondly microcalcifications were enhanced by utilizing Laplacian operator in spatial domain. Thirdly Butterworth highpass filters was adopted to located microcalcifications incipiently. Lastly two-dimensional entropic thresholding was applied to segmentation image, thus microcalcifications were detected and were marked in mammograms. The proposed method could avoid misdiagnosis because of doctors’ fatigue causing by reading large amount of films. It played an important role of assisting doctors to diagnose and it also improved the efficiency and accuracy of diagnosis.