Contrast-to-noise ratios of liver lesions using subtraction imaging on multiphase 64-detector row CT

To measure contrast-to-noise ratios of liver lesions on conventional enhanced and digitally subtracted multidetector row computed tomography (CT) images.

This study was approved by our hospital internal review board (IRB) and all collected data were evaluated in a Health Insurance Portability and Accountability Act (HIPAA)-compliant manner. Subtracted datasets, using pixel-by-pixel subtraction of the post-contrast images from the pre-contrast images, were created from the 64 detector-row CT of patients undergoing three-phase examination of the liver (unenhanced, arterial phase, and portal venous phase). Regions of interest were used to calculate the contrast-to-noise ratios between the lesions and the background liver parenchyma on both the post-contrast and subtracted datasets using the following formula: (Lesion mean (HU) - Liver mean (HU))/standard deviation of mean outside patient (HU). These ratios were compared using a mixed linear statistical model.

Contrast-to-noise ratios were calculated for 64 lesions in 50 consecutive patients. Of the 64 lesions, 42 were hypervascular and 22 were hypovascular. Subtracted datasets yielded statistically significant higher contrast-to-noise ratios of hypervascular lesions compared to normal liver parenchyma (p<0.0001). Subtraction did not yield a statistically significant improvement in contrast-to-noise ratios for hypovascular liver lesions (p=0.16).

Post-processed subtraction CT images generate increased contrast-to-noise ratios for hypervascular liver lesions. As this technique is easy to perform and does not involve additional radiation exposure, it should be considered when evaluating for suspected hypervascular lesions.