This paper proposes an automated extraction method for liver regions in three dimensional abdominal CT images. The method consists of two steps; a rough extraction step of liver regions using original and triple-phase helical CT images and a precise extraction step which can refine the extracted region by the first step. The latter step uses a level set method whose speed function is designed based on knowledge of shape and position of liver regions. This paper also shows expeimental results of applying the proposed method to fifty two abdominal CT images and evaluating the performances.

View full abstract