Effective dose reduction in dual-energy flat panel x-ray imaging: technique and clinical evaluation

Dual-energy (DE) chest radiography with a digital flat panel (DFP) shows significant potential for increased sensitivity and specificity of pulmonary nodule detection. DFP-based DE produces significantly better image quality compared to Computed Radiography (CR) due to high detective quantum efficiency (DQE) and wide energy separation. We developed novel noise reduction filtering that significantly improves image quality at a given dose level, thereby allowing considerable additional dose reduction compared to CR. The algorithm segments images into structures, which are processed using anisotropic smoothing and sharpening, and non-structures, which are processed using isotropic smoothing. A fraction of the original image is blended with the processed image to obtain an image with improved noise characteristics. DE decomposed radiographs were obtained at film equivalent of 400 speed chest exam dose for 12 patients (set A) and at twice the dose for 7 other patients (set C). Images from set A were filtered using our algorithm to form set B. Images were evaluated by four radiologists using a noise rating scale. A two-sample t-test showed no significant difference in ratings between B and C, while significant differences were found between A and B, and A and C. Therefore, our algorithm enables effective patient dose reduction while maintaining perceptual image quality.