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 Consultant, INFINITT Healthcare Co, Ltd

Our method can be used to differentiating malignant and benign nodules by calculating the ratio of GGO and solid areas.

The ground glass opacity (GGO) proportion in pulmonary nodule is one of the most important prognostic factor because nodules with pure and mixed-GGO are more likely to be malignant than that with solid opacity. However, it is difficult to accurately segment GGO nodule and separate GGO and solid areas due to its indistinct boundary and vessels inside nodule. In this paper, we propose an automatic GGO nodule segmentation method in chest CT images using improved multi-phase deformable model.

Our histogram modeling reduces the positional sensitivity of initial contour in multi-phase deformable model. Our improved multi-phase deformable model extracts GGO and solid areas simultaneoulsy. Our vessel removal filtering helps to separate vessels indide nodule from solid areas of GGO nodule and accurately calculate the ratio of GGO and solid areas.

The CT images of thirty patients with pure and mixed-GGO nodules were obtained on the Lightspeed Ultra CT scanner (GE) and the Sensation 16 Scanner (Simens). Each image had a matrix size of 512 x 512 pixels with in-plain resolutions ranging from 0.56 to 0.79 mm. The slice thickness ranged from 1.0 to1.25 mm and the number of images per scan ranged from 192 to 473. Optimal volume circumscribing a nodule is decided by clicking inside of GGO nodule. To decide an appropriate threshold value of solid areas in GGO nodule and localize an initial contour, histogram modeling is performed by Gaussian Mixture Modeling with three components including lung parenchyma, nodule, and chest wall or vessels. To segment GGO and solid areas simultaneously, multi-phase deformable model is performed by chan-vese approach. To prevent vessels inside nodule from estimating as a solid area, vessel-like structures inside GGO nodule are removed by Hessian-based filtering. To evaluate the performance of proposed method, segmentation results were visually assessed and compared with traditional multi-phase deformable model.  

Jung, J, Hong, H, Goo, J, Automatic Segmentation of Ground Glass Opacity Nodule in Chest CT Images Using Improved Multi-Phase Deformable Model.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12036250.html