High-spatial-frequency (bone) algorithm improves quality of standard CT of the thorax.

The high-spatial-frequency (bone) reconstruction algorithm has been shown to be superior to the standard algorithm in the assessment of thin-section images of the lung parenchyma. We compared the high-spatial-frequency and standard algorithms in the evaluation of normal and abnormal lung parenchyma and mediastinum on conventional 10-mm-collimation CT scans. Three observers, blinded to the reconstruction algorithm used, compared both algorithms on identical images in 31 patients with a variety of diseases. Lung parenchymal detail was improved with the bone algorithm in 94% of scans (p less than .001). The definition of mediastinal contents with the bone algorithm was judged equal or superior to the standard algorithm in 94% of images (p less than .001). Scanning of a line-pair-resolution phantom demonstrated a 28% improvement in image resolution with the bone algorithm. Although the bone algorithm resulted in an increase in visible noise, the overall visual quality of the images was considered equal to or greater than that of those reconstructed on the standard algorithm in 100% of parenchymal (p less than .001) and 85% of mediastinal (p less than .001) images. We conclude that routine use of the bone algorithm results in improved spatial resolution and definition of pulmonary parenchymal detail without appreciably degrading the overall visual quality of mediastinal images.