Jangpyo Bae MS, Abstract Co-Author: Nothing to Disclose

Namkug Kim PhD, Presenter: Nothing to Disclose

Joon Beom Seo MD, PhD, Abstract Co-Author: Nothing to Disclose

Sang Min Lee MD, Abstract Co-Author: Nothing to Disclose

The accurate delineation of the thoracic cage region is vital for the various kinds of clinical applications including thoracic cage volumetry and mediastinum fat quantification of patients with chronic obstructive pulmonary disease (COPD). The purpose of this study is to develop and evaluate an automatic segmentation method for the thoracic cage.

Volumetric CT scans of fifty one patients with chronic obstructive pulmonary disease (COPD) were performed by a 16-multi detector row CT scanner (Siemens Sensation 16) with in 0.75mm collimation. The thoracic cage region was separated from the other region by using the inner thoracic wall and the diaphragm surface by using a 3D surface fitting method. The inner thoracic wall can be composed of 4 exclusive quarter-surfaces of 3D closed thoracic cage made from threshold-based rib segmentation with the ray projection. In the case of diaphragm, the lower surface of each lung was used for input of the 3D surface fitting. Therefore, thoracic cage region was calculated from these five surfaces. Because accuracy of diaphragm surface is low, the supplementary segmentations of the heart and the surrounding fat of that heart were performed by a sphere shape prior level set and a typical level set method respectively with two manual points on the top and bottom of the heart. To assess the accuracy of the proposed algorithm, the segmentation results of 51 patients were compared with those of manual segmentation by an expert thoracic radiologist. Evaluation metrics for segmentation accuracy include volumetric overlap ratio (VOR), false positive ratio on VOR (FPRV), false negative ratio on VOR (FNRV), average symmetric absolute surface distance (ASASD), average symmetric squared surface distance (ASSSD), and maximum symmetric surface distance (MSSD). 

Mean and SD of VOR, FPRV, FNRV, ASASD, ASSSD, and MSSD were 94.09±2.10%, 1.88±1.84%, 4.03±1.93%, 1.15±0.54 mm, 4.18±1.59 mm, and 42.56±13.12 mm, respectively. 

We proposed the automatic thoracic cage segmentation method with rib, thoracic wall, diaphragm, and heart segmentation with 3D surface fitting in the 3D volumetric CT data, which might be clinically applicable.

Our method would be useful in the various kinds of clinical applications including thoracic cage volumetry, and mediastinum fat quantification of patients with chronic obstructive pulmonary disease.

Bae, J, Kim, N, Seo, J, Lee, S, Automatic Segmentation of the Thoracic Cage Using Rib and in the Volumetric CT Data .  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13044453.html