This paper proposes a new method for extracting liver regions from 3D abdominal X-ray CT images without contrast medium. The previous method using the thresholding and the distance transformation cannot extract thin regions or regions at low CT value, such as disease region. So the proposed method extracts rough liver regions at first. Then, more detailed regions are extracted by using distribution feature of CT value and the information of the position. In this time, target regions are limited by using rib cage that the shape almost unchange, overextraction is restrained. In addition, our method extracts disease region by using morphological operation. We applied the proposed method to 14 cases of abdominal CT images taken without using contrast medium. The experimental results showed that the proposed method was significantly improved in extraction.

View full abstract