the early diagnosis of lesions in mammography would decrease the rate of breast cancer mortality. In this research an algorithm for detection of microcalcification clusters in digital mammograms based on special features extraction and multi-resolution analysis was proposed. The algorithm consists of four main parts. Segmentation algorithm separates the breast region from the background Image by using a new histogram thresholding method. It reduces the background noise effect by focusing the detection algorithms on breast region. The Second part is an initial detection algorithm that is applied to the Region of Interest (ROI) to separate uncertain parts in which microcalcifications may be present. By using this algorithm, the size of ROI was reduced. So the execution time and false Breast cancer is the most common form of cancer and the main reason for the mortality caused by cancers among women. Mammography is the most effective procedure for an early diagnosis of the breast cancer. Due to low doze X-ray Imaging of breast soft tissue, the contrast in mammograms, especially in dense breasts, is low. Detection of microcalcifications, which are one of the main features considered by radiologists for diagnosis, is so difficult, especially in the first stages of forming. According to physicians, positive rate will decrease in the main detection algorithm. In the main detection algorithm, a multi-resolution analysis by means of the wavelet packet transform (WPT) for detecting microcalcifications is exploited. Noise reduction and microcalcification detection is performed using WPT coefficients analysis. Finally, a clustering algorithm based on morphological methods was proposed that can separate clustered microcalcifications from single ones. From radiography point of view, clustered microcalcifications are more important than scattered ones. The final results show that using WPT instead of discrete wavelet transform improves the detection algorithm and reduces the false detections rate.