Detection of abnormalities in chest radiographs is limited by the complex and variable nature of the normal background anatomical structures. Digital radiography allows various types of image processing to be performed, but techniques that improve the visibility of abnormal findings also tend to emphasize certain features of normal anatomy. Ideally, the visibility of pathological findings would be improved selectively, while normal anatomical structures would be suppressed. In the case of patients who have had a previous chest radiograph, an opportunity exists to enhance selectively areas of interval change including regions with new or altered pathology, by using the previous radiograph as a subtraction mask. The applicants propose to develop and evaluate a technique to produce a digital difference image from sequential chest radiographs. It is anticipated that this will improve detection of various types of interval change by suppressing the background anatomy while enhancing new or altered pathology. The clinical utility of the difference imaging technique will be evaluated with observer tests using ROC analysis. Using previously developed algorithms, the midline and rib-cage outlines will be identified after density and contrast normalization. A non-linear warping technique will be applied to one image to correct for differences in projection, and patient positioning will be standardized using lasers and simple positioning devices in order to minimize misregistration. Preliminary experience with over 300 patients, and the results of an observer performance test, strongly suggest that this technique improves diagnostic accuracy for a wide range of pathology including lung cancer and pulmonary infections.
