The Effect of Incorrect HU on Standardized Uptake values

Objectives Contrast enhanced computed tomography converts Hounsfield units (HU) into Positron Emission Tomography (PET) attenuation correction (AC) factors (1). On a PET/CT system, standardize uptake values (SUV) are used to determine maximum ¹⁸F FDG activity within a region of interest (ROI) (2). Comparing serial SUV values provides a quantitative measurement of cancer therapy effectiveness within the ROI. Generally, a decrease of 25 percent in the SUV indicates a positive response to therapy. Failure to indicate whether iodinated IV contrast was used prior to PET images can affect the SUV (3). This is important for head and neck cancers, where contrast may be in close proximity to tumors. The purpose of this study determined how patient data is affected if the technologist omits indicating that contrast was given.

Methods Raw data were obtained using a GE Discovery 710 PET/CT system. 31 patients were been given IV CT contrast immediately prior to a PET scan. These data were reconstructed with HU indicative of IV contrast versus the HU of blood vessels. A ROI was drawn and copied onto the renal pelvis of each image and the SUV_max was recorded.

Results The data with incorrect HU values showed a SUV of 4.06 percent greater than the calculated SUV_max of the data with the correct HU value. The mean SUV_max percent difference was 4.06 for the incorrect AC values assigned to the blood vessels. The median percent difference was 1.73 percent, with a range of 0.24-18.13 percent.

Conclusions A twenty-five percent reduction in SUV_max is viewed as a good response to therapy. This study determined how the patients' SUV_max was changed by the incorrect HU values when using IV contrast. In all of the patients the ROI with incorrect HU values showed an increase in SUV_max when compared to those with the correct HU values. Over AC due to the incorrect HU consistently affected the SUV_max; however, none of the patient’s SUV_max were altered by more than 25 percent. The SUV_max error alone would not affect the patient’s outcome. In the future, using a larger patient database may improve the statistical relevance of this study.