Ju Lip Jung BEng, Presenter: Nothing to Disclose

Helen Hong PhD, Abstract Co-Author: Nothing to Disclose

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

Kyunghee Lee MD, Abstract Co-Author: Nothing to Disclose

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

Jae Yeon Wi MD, Abstract Co-Author: Nothing to Disclose

The proposed method can be used to differentiate malignant and benign nodules by analyzing the volumetry changes of GGO and solid component in follow-up chest CT scans.

To differentiate malignant and benign nodules through computer-aided volumetry of ground-glass opacity (GGO) and solid component, we propose a GGO nodule segmentation method using asymmetric multi-phase deformable model with intensity constraint and vascular structure elimination.

The chest CT images were obtained on the Lightspeed Ultra CT scanner (GE) and the Sensation 16 Scanner (Siemens) with various reconstruction kernels (B30f, B50f and B60f). The dataset is consisted of 10 pure GGO nodules and 24 mixed-GGO nodules (diameter 7.4-25.7mm, mean diameter 17.1±5.1mm). To extract initial GGO and solid component, optimal volume circumscribing a nodule was decided by clicking inside nodule and solid component was extracted by applying thresholding with -200HU. Then GGO was extracted by estimating the adaptive threshold value based on intensity histogram modeling. To segment final GGO and solid component, GGO and solid component were simultaneously separated from lung parenchyma using asymmetric multi-phase deformable model with intensity constraint. To eliminate vessels inside GGO nodule, vessel-like structures are enhanced by Hessian-based vessel enhancement filtering with oval blob-like structures suppression. To evaluate the performance of computer-aided volumetry, solid component proportion difference (SCPD) between computer-aided volumetry and manual volumetry was measured. The solid component proportion was calculated as (solid component volume / GGO nodule volume) and the SCPD was calculated as ((computer-aided volumetry – manual volumetry) / manual volumetry × 100). The average SCPD was 2.7±6.6% and the limits of agreement were 15.7% and -10.4%.

Our asymmetric multi-phase deformable model with intensity constraint accurately seperates GGO and solid cpmponent from lung parenchyma. Our vessel enhancement filtering with oval blob-like structures suppression helps to eliminate vessels without the loss of solid component and accurately measure the GGO and solid component volumes.

Jung, J, Hong, H, Goo, J, Lee, K, Park, S, Wi, J, Computer-aided Volumetry of Ground-glass Opacity and Solid Component through the Nodule Segmentation and Vascular Structure Elimination in Chest CT Images .  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13021200.html