Won Chang MD, Presenter: Nothing to Disclose

Chang Min Park MD, PhD, Abstract Co-Author: Nothing to Disclose

Sang Joon Park, Abstract Co-Author: Nothing to Disclose

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

Jin Mo Goo MD, PhD, Abstract Co-Author: Research Grant, Guerbet SA Research Grant, Toshiba Corporation

To retrospectively investigate the differentiating value of computerized three-dimensional (3D) shape analysis between pulmonary benign and malignant nodules

Between January 2010 and June 2012, we identified 113 patients (59 men and 54 women; mean age, 58.7 ± 13.0 years) with 113 pathologically-confirmed pulmonary nodules ≤ 2cm in size (mean size, 1.45 ± 0.40 cm; 62 malignant and 51 benign nodules) on thin-section chest CT. Each lung nodule was manually-segmented from the surrounding lung parenchyma on axial CT images and 3D shape features of each nodule were calculated using an in-house software program. To evaluate the differentiating value of these 3D shape features between benign and malignant nodules, comparison statistics and receiver-operating characteristics curve (ROC) analysis was performed.

Between benign and malignant nodules, there were significant differences in nodule’s sphericity, discrete compactness and 3D roundness. Compared with malignant nodules, benign nodules showed significantly higher sphericity (0.767 vs. 0.653, p<0.001), higher discrete compactness (0.572 vs. 0.449, p<0.001) and 3D roundness (0.711 vs. 0.686, p=0.013). Area under ROC curves (AUCs) of sphericity, discrete compactness and 3D roundness in discriminating benign from malignant nodules were 0.803, 0.721 and 0.604, respectively. As for sphericity, highest Youden index was obtained using a cut-off value of 0.734 (sensitivity, 66.7%; specificity, 82.3%), and benign pulmonary nodules were diagnosed with 100% specificity when using a cut-off value of 0.839 (sensitivity 31.4%).

Computerized 3D shape analysis such as nodule’s sphericity has a potential as a differentiating tool between pulmonary benign from malignant ones.

(dealing with thin section chest CT); Computerized 3D shape analysis of lung nodules can differentiate benign from malignant ones ;and is recommended as part of initial evaluation prior to the biopsy.

Chang, W, Park, C, Park, S, Lee, S, Goo, J, Discrimination of Pulmonary Benign from Malignant Nodules Using a Computerized Three-dimensional Shape Analysis.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13027335.html