Abstract: When existing a relatively large juxta-pleural nodule in lung CT image，because of the lacunas in the segmented lung，the attached nodules may not be distinguishable from the lung.However，since there exist some similarities between the lung’s flank outline area and the rib contours，we can find the growth points of suspicious nodules more precisely by using cubic spline to fit the rib contours at first，then apply regional growth to the combined lung contour image and the rib contour image.The experiment result shows this algorithm can cut the juxta-pleural lung nodules from the CT chest image more effectively.

Key words: CT image, lung area, suspicious juxta-pleural lung nodule, rib contour, spline fitting

摘要： 在计算机辅助CT图像肺部肿瘤诊断系统中，当肺区内部存在较大粘连肿瘤时，将导致分割出来的肺区存在较大的缺陷，因而在肺区内部就无法识别粘连肿瘤。利用肋骨轮廓与肺区轮廓形状相似的特性，先用三次立方样条曲线拟合出肋骨的内部轮廓，然后根据肋骨轮廓图与肺区轮廓图找出可疑粘连肿瘤的种子点，最后将肋骨轮廓图与肺区图相结合进行区域生长找到疑似粘连肿瘤。实验结果表明该方法能有效的从CT图像中分割出疑似粘连肿瘤。

关键词: CT图像, 肺区, 疑似粘连肿瘤, 肋骨轮廓, 样条拟合