%0 Journal Article
%T Automatic Segmentation of Lung Carcinoma Using 3D Texture Features in 18-FDG PET/CT
%A Daniel Markel
%A Curtis Caldwell
%A Hamideh Alasti
%A Hany Soliman
%A Yee Ung
%A Justin Lee
%A Alexander Sun
%J International Journal of Molecular Imaging
%D 2013
%I Hindawi Publishing Corporation
%R 10.1155/2013/980769
%X Target definition is the largest source of geometric uncertainty in radiation therapy. This is partly due to a lack of contrast between tumor and healthy soft tissue for computed tomography (CT) and due to blurriness, lower spatial resolution, and lack of a truly quantitative unit for positron emission tomography (PET). First-, second-, and higher-order statistics, Tamura, and structural features were characterized for PET and CT images of lung carcinoma and organs of the thorax. A combined decision tree (DT) with K-nearest neighbours (KNN) classifiers as nodes containing combinations of 3 features were trained and used for segmentation of the gross tumor volume. This approach was validated for 31 patients from two separate institutions and scanners. The results were compared with thresholding approaches, the fuzzy clustering method, the 3-level fuzzy locally adaptive Bayesian algorithm, the multivalued level set algorithm, and a single KNN using Hounsfield units and standard uptake value. The results showed the DTKNN classifier had the highest sensitivity of 73.9%, second highest average Dice coefficient of 0.607, and a specificity of 99.2% for classifying voxels when using a probabilistic ground truth provided by simultaneous truth and performance level estimation using contours drawn by 3 trained physicians. 1. Introduction The clinical outcome of radiation therapy is closely linked to the ability to deliver dose within tightly confined boundaries to maximize target control while minimizing dose to surrounding tissue to reduce the probability of complications. This is the basis for treatment planning in radiotherapy, the success of which relies on the minimization of geometric and dosimetric uncertainties. Target uncertainty is compounded by movement when disease is present in the thorax which can be minimized with the aid of breath hold techniques such as audio coaching or active breathing control using a frame. However, for many patients who suffer from breathing difficulties, these are often not tolerable solutions. Fortunately, a host of recent technological developments and techniques have aided in reducing the errors at each stage of treatment from simulation, planning, quality assurance, and delivery [1]. Modern delivery techniques can conform dose fields geometrically to within 2£żmm and dosimetrically to within 2-3% [2]. However, this is largely undermined by uncertainty in target definition. Interobserver variability in target segmentation is one of the largest sources of error in radiotherapy [3], with ratios of the largest to smallest
%U http://www.hindawi.com/journals/ijmi/2013/980769/