Chest radiography is the main screening tool for the detection of asymptomatic lung disorders including cancer. However, due to the variations in tissue thickness over the chest, the radiograph is characterized by large variation in photon fluence between the thinner lung ara and remaining thick sections. By conventional means this variation falls outside the useful exposure range of radiograhic film and produces images that are severely underexposed over the thicker body parts. While digital chest radiography offers the capability to acquire and display projection data over all body parts due to their use of large dynamic range detectors it is still subject to image signal/noise degradation from the produced photon fluence over the thicker body parts. This nonuniform signal/noise ratio will limit the low contrast detectability as well as the utility of image processing and dual energy algorithms to enhance chest radiography. We have recently developed an automatic scanning """"""""equalization"""""""" technique for film radiography to regionally adjust the patient exposure in response to the patients transmission distribution so as to avoid large excursions in film exposure. We will apply this equalization technique to digital chest radiography to minimize the detected photon fluence variations and therby improve the uniformity of the image signal/noise distribution. We will implement the scanning technique to a prototype digital chest radiography system and explore the significance of regional image noise control by physical measurement and low contrast detectability studies. We will explore a new technique in improving image quality through two approaches at numerically compensating the noise degrading effects of scatter contamination. The impact of these improvements will be tested on candidate image processing techniques. The combination of improved image noise and enhanced scatter compensation may provide a means of optimizing chest radiography with significant benefits on diagnostic accuracy over conventional methods. Evaluation of this technique will be made by measurements of physical parameters of image quality, low contrast perception studies and a thorough clinical evaluation.
