Active contour bilateral filter for breast lesions segmentation on ultrasound images

Abstract

Breast cancer is most malignancy in woman worldwide. Mammography, the primary modality for early detection of breast cancer in women > 40 years old has some limitations. The low sensitivity of Mammography to find abnormalities in dense breast and young woman and the low specificity to determine the cystic and solid lesions causes unnecessary biopsy. In recent decades, ultrasound imaging has been widely used as the best modality in identifying breast lesion and it has proven to be a valuable addition to mammography. Ultrasound can distinguish between cystic and solid mass thus improving the accuracy of diagnosis. The crucial step in this process is lesion segmentation because a lot of important features for identifying suspicious lesions are related to lesion shape and boundary. Clinicians are expecting the lesion area can be localized from normal tissue appropriately. Active Contour technique is proposed for this purpose. It can detect the edges of breast lesions and separating normal tissue iteratively. But the character of the ultrasound image that brings speckle noise make difficult to do segmentation process. It also resulted in interpretation errors and inaccuracies diagnosis made by a doctor. Therefore speckle noise reduction method is needed in order to become more assertive edge of the lesion and well segmented without losing information inside the image. Here, speckle noise reduction is done with Bilateral Filter. Comparison of Active Contour breast lesions segmentation with and without the Bilateral Filter is attached at the end of this paper. The combination of Bilateral Filter with Active Contour showed better results with the edge of the lesion is firmly and clear.