Automatic Rib Positioning Method in CT Images

Efficiently, accurately and automatically positioning the ribs in CT images and giving tips of which rib it is on cross-sectional plane will be very helpful for doctor's clinical diagnosis. This paper presented a new method of automatic rib positioning and extraction in CT images. According to the anatomical characteristics of ribs, the first rib will presents as a sloping line gradually when we slice the sagittal plane images from the subcentral position of the body to the right or left side, while the other ribs will present as small ovals at that time. Based on this feature, we did the Hough transform on a series of sagittal planes and performed an automatic detection to find the first rib. When finding the first rib, the corresponding plane was defined as the Key Saggital Plane (KSP), and we positioned the ribs and defined the initial seed point of each rib on this plane. The rib extraction was a kind of tracing-based one using the initial seeds defined above. We evaluated our method on 21 chest CT data sets, including 11 cases of male and 10 cases of female. The age range was from 18 to 90, among which 12 cases were over the age of 55. Besides, the osteoporosis patients were involved. We tested our algorithms on the PC with XP-Pro/SP2, P4-3.0GHz, 2.0GB Memory. It was proved that the average number of the extracted ribs is 11.5 of each side, and the positioning was correct. From the efficiency point of view, the average number of image slices of one patient was 104, and the running time of the algorithm was 4.95 seconds.