Abstract
90
Objectives The standard method for estimating lung shunt magnitude before 90Y selective internal radiotherapy (SIRT) involves computing conjugate lung & liver counts from anterior & posterior views, but ignores background (Bkgd) counts present in these images. This investigation sought to determine which scintigraphic method is most accurate for computing shunt values.
Methods Saline bag phantoms simulating 1 1000 mL-liver, 2 50 mL-liver lesions, & 2 500 mL-lungs, submerged in a 6,000 mL cylindrical water phantom were used. Data were examined retrospectively for 25 pts to derive ranges of counts & Bkgd-to-lung (Bkgd:L) ratios characteristic of clinical SIRT images. Based on these data, 99mTc activity was adjusted to produce 5 simulations with no Bkgd and 9 simulations with Bkgd:L ranging from 19%-79%, for simulated shunt values ranging from 8%-25%. Simultaneous anterior & posterior conjugate views & SPECT were performed using standard clinical acquisition protocols. Three methods were used to compute shunt ratios: Method 1 (manufacturer suggested): geometric mean of counts within a single anterior & posterior lung region compared to conjugate liver counts; Method 2: background counts per pixel subtracted from counts of Method 1; Method 3: tomograms reconstructed using manufacturer’s resolution recovery algorithms & attenuation correction (Chang method).
Results With no Bkgd, all methods agreed equally well with true values (ANOVA p=0.88). With Bkgd, Method 1 overestimated percent shunting (+5.3±2.2%, p<0.0001), Method 2 underestimated percent shunting (-6.9±4.2%, p=0.0013) & Method 3 overestimated shunting (+1.0±1.0%, p=0.03). Pearson correlation coefficients vs. true shunt values were r=0.93, 0.70 & 0.99 for Methods 1, 2 & 3. Errors were significantly correlated to increasing Bkgd levels for Methods 1 & 2 (p=0.003 & 0.001), but not Method 3 (p=0.35).
Conclusions With background activity the conventional conjugate view method overestimates shunting while attempted background correction underestimates shunting