Purpose: To investigate the impact of flexible noise control (FNC) processing (Fujifilm, Tokyo, Japan) on digital radiography chest imaging.
Methods: An anthropomorphic chest phantom was imaged using a clinical chest protocol with 85 kV and five radiation dose levels (0.7, 1.0, 1.4, 1.8, and 2.2 mAs) at 40’’ source-to-image distance with scatter removal image processing applied. All images were processed without and with FNC. Selected options for the four processing parameters of FNC included filter control FFC (C and F), frequency balance FNB (A, C, E, and G), density control FNT (A, B, and C), and strength control FNE (0.1, 0.5, and 0.9), resulting in 72 combinations. Subtractions were performed between images without and with FNC from the same setting to assess noise and grayscale changes. Noise-only images were generated by subtracting two identical acquisitions and two regions-of-interest (ROIs), one over lung field and the other over liver area, were drawn to quantify the noise level for each setting. Texture analysis was performed in the liver ROI using the gray-level co-occurrence matrix (GLCM). Edge-preserving capability was evaluated using a line profile through a lung/rib interface.
Results: Use of FNC resulted in 15-70% noise reduction depending on the dose levels, parameter choices, and ROI selections. Options in FNB and FNE parameters did not show performance difference. Option F for FFC and A for FNT demonstrated the highest noise reduction compared to other options. Information measure of correlation from GLCM demonstrated the largest difference among the parameter combinations, with CGC the closest to baseline and FGA/B the most different. The appearance of edge interfaces was also affected, with FGB and CGB providing better edge-preserving capabilities.
Conclusion: The impact of FNC on chest imaging was demonstrated using a phantom. Noise suppression and edge-preserving capability depend on radiation dose level and FNC parameter choice.