In this paper, we describe a method for automated extraction of abdominal organs from uncontrasted 3D abdominal X-ray CT images based on anatomical knowledge. In abdominal CT image, intensity contrasts between abdominal organs are low and they exist very closely. Therefore, any extraction methods based on CT values, such as a region-growing method, cannot segment each organ correctly. On the other hand, shapes, positions and distributions of CT values of organs are known. The proposed method extracts each organ by a region-growing method, which uses not only CT values of organs but also anatomical knowledge mentioned above. Before extracting each organ, the processing area for each organ is defined by using anatomical knowledge of its shape and position so as to make the extraction process stable. In the extraction process, distribution information of CT values of each organ is used as a condition of the region-growing. The proposed method is applied to fourteen cases of 3D abdominal X-ray CT images. The extraction results showed that each abdominal organ can be extracted satisfactorily and stably.

View full abstract