Tumor of 2D Mammography Mapping in Automated Breast Ultrasound

• Main Authors: Pei-Fan Lin, 林倍帆
• Other Authors: 張瑞峰
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/19098445325109641329

碩士 === 國立臺灣大學 === 資訊工程學研究所 === 105 === Mammogram and ultrasound (US) play an important role in breast cancer diagnosis. Nowadays, Automated Breast Ultrasound (ABUS) has been developed to overcome the limitation of conventional handheld US such as operator dependence and time-consuming in ultrasound screening. In the present clinical workflow, the radiologist needs to review both mammography and ABUS image for checking a suspicious tumor. However, this is a time-consuming process. To increase the efficiency of the screening process, we develop a tumor mapping method between mammogram and ABUS image to identify the same tumor location in mammogram and ABUS image in a more systematic way. Our method involves the following steps to determine the location of the tumors. We first measure the border of a breast in the mammogram image for getting the projection of the tumor on the breast surface. Once we get the projection of the tumor on the breast surface, we use it with nipple and chest wall as reference points to determine the tumor location relative to the nipple in mammogram and ABUS image. Given the relative location of the tumor to the nipple in mammogram, we can use it to extract the actual tumor location from the ABUS image. In this thesis, we use 44 samples to evaluate our proposed method. Each case consists of MLO and CC view mammograms and ABUS image. Our estimation provides the accuracy of the tumor locations within 2cm to the actual location in both mammogram and ABUS image.