ABSTRACT
The purpose of this study was to develop and validate a technique for three-dimensional (3D) modelling of small lung nodules on paediatric multidetector array computed tomography (MDCT) images. Clinical images were selected from 21 patients (<18 years old) who underwent MDCT examinations. Sixteen of the patients had one or more real lung nodules with diameters between 2.5 and 6 mm. A mathematical simulation technique was developed to emulate the 3D characteristics of the real nodules. To validate this technique, MDCT images of 34 real nodules and 55 simulated nodules were randomised and rated independently by four experienced paediatric radiologists on a continuous scale of appearance between 0 (definitely not real) and 100 (definitely real). Receiver operating characteristic (ROC) analysis, t-test, and equivalence test were performed to assess the radiologists' ability to distinguish between simulated and real nodules. The two types of nodules were also compared in terms of measured shape and contrast profile irregularities. The areas under the ROC curves were 0.59, 0.60, 0.40, and 0.63 for the four observers. Mean score differences between simulated and real nodules were -8, -11, 13, and -4 for the four observers with p-values of 0.17, 0.06, 0.17, and 0.26, respectively. The simulated and real nodules were perceptually equivalent and had comparable shape and contrast profile irregularities. In conclusion, mathematical simulation is a feasible technique for creating realistic small lung nodules on paediatric MDCT images.
Subject(s)
Algorithms , Computer Simulation , Imaging, Three-Dimensional , Lung Neoplasms/diagnostic imaging , Solitary Pulmonary Nodule/diagnostic imaging , Tomography, X-Ray Computed/methods , Child , Humans , ROC Curve , Reproducibility of Results , Sarcoma/diagnostic imagingABSTRACT
We compared the use of barium enema (BE) and colonoscopy in evaluating patients with chronic gastrointestinal tract bleeding by dividing into three groups 329 consecutive colonoscopies and 207 consecutive BE examinations done with chronic gastrointestinal tract bleeding as an indication. In the first group, of 96 patients with negative results of BE studies, subsequent colonoscopy showed carcinoma of the colon in 16%, polyps larger than 1 cm in 21%, and other causes in 20%. In 43% the colonoscopy gave negative results or was incomplete. In the other two groups we directly compared findings of the 207 BE and the 233 remaining colonoscopies when each was used as a primary diagnostic test. Colonoscopy was found to have fewer negative results (74% vs 43%), fewer inconclusive examinations requiring repeat (19% vs 3%), and more positive correct findings to explain the cause of bleeding (54% vs 5%).
