Effects of intravenous CT contrast on attenuation correction of non-bone ^99mTc SPECT/CT

Objectives To quantify errors in non-bone ^99mTc SPECT/CT caused by presence of intravenous (IV) contrast in CT images used for attenuation correction (AC), since the magnitude of AC is greater at 140 keV (SPECT) than 511 keV (PET) but the impact of IV contrast on CT-AC of SPECT is not well characterized.

Methods SPECT/CT scans of a 60-cc syringe containing ^99mTc solution inserted in an e-density phantom were acquired at 4 different activities. Additional co-registered CT scans at different kVp’s were performed with the ^99mTc solution replaced with 7 different IV contrast concentrations (0.64, 3.2, 6.4, 12.8, 20, 32, 64 mg/mL). The linear attenuation coefficient (LAC) of contrast solutions were calculated using NIST XCOM software to derive the modified bi-linear CT-to-LAC transform accounting for the presence of contrast for CT-AC. The relative change (bias) in SPECT signal was calculated when CT-AC images contained IV contrast using the standard and modified CT-to-LAC transforms.

Results The increase in CT number (HU) was proportional to the contrast concentration (17.1 HU/mg/mL at 130 kVp). The magnitude of proportionality was inversely related to the CT kVp. The difference between the true LAC for IV contrast and that using CT-AC map increased linearly with its concentration. The errors in LAC at 140 keV was <10% at contrast concentration <6.4 mg/mL (1:50 dilution) but increased to 35% at contrast concentration of 32 mg/mL (1:10 dilution). The CT-to-LAC transform in the presence of contrast substantially flattened above 0 HU; the slope reduced from 1.37E-3 to 2.66E-4. The bias in reconstructed SPECT signal was independent of the ^99mTc activity but directly proportional to the contrast concentration in the CT image (0.35 %/mg/mL at 140 keV). The ^99mTc SPECT bias was ~11% for contrast concentration of 32 mg/mL but decreased to <2% using the modified CT-to-LAC transform that accounted for IV contrast.

Conclusions IV contrast in CT attenuation maps biases the SPECT signal in proportion to its concentration. However, the ^99mTc SPECT bias with typical IV contrast administrations (ΔHU<500) is estimated to be <10%