The purpose of these studies is to evaluate whether digital tomosynthesis can serve as an adjunct to conventional chest radiography for improving clinical detection of pulmonary nodules. Digital tomosynthesis augments a conventional chest exam by providing three-dimensional information through a series of longitudinal slice images. These slice images are reconstructed from a set of discrete projection images acquired at different angles, using a conventional x-ray tube and a new digital flat-panel x-ray detector. The tomosynthesized images may be viewed slice-by-slice or as a 3-D volume-rendered projection in a """"""""virtual fluoroscopy'' viewing environment. The diagnostic benefit of tomosynthesis is the use of 3-D information to improve detection and discrimination of pulmonary nodules by eliminating the confusion of overlying structures. CT would remain the gold standard for clinical workup once nodules are detected, with tomosynthesis providing a low-dose/low-cost method for improving initial detection accuracy. Clinically the tomosynthesis images would be acquired whenever a digital PNlateral chest exam is taken, but with reduced exposure for the lateral image such that the total patient dose from the PA/lateral/tomosynthesis exam would be about comparable to the overall dose from a conventional PNlateral chest exam. It is anticipated that tomosynthesis will enable higher sensitivity and specificity for pulmonary nodule detection than conventional chest radiography alone, and hence will potentially lead to earlier detection of lung cancer, at a cost significantly less than low-dose CT. This proposal is a competing continuation of a previous study that developed and optimized the tomosynthesis technique. The present study will conduct a clinical evaluation of tomosynthesis for pulmonary nodule detection in a cohort of human subjects with CT-proven nodules. This study will have three parts: (1) an evaluation of tomosynthesis using radiologists as observers, (2) development and evaluation of a computer-aided diagnosis (CAD) method using three-dimensional tomosynthesis data for pulmonary nodule detection, and (3) a measurement of the accuracy with which radiologists can use tomosynthesis to localize the depth and diameter of nodules.
The specific aims of the project also include pilot studies to optimize the tomosynthesis method before the large-scale collection of human subject data. These pilot studies will characterize and minimize effects of patient motion, optimize the number of reconstructed planes and patient entrance exposure, and evaluate slice viewing vs volume rendering.
Showing the most recent 10 out of 20 publications
