Room: Exhibit Hall
Purpose: Pre-treatment Tc-99m MAA (macroaggregated albumin) scans are commonly used for Y-90 microsphere treatment planning. Here we investigate the accuracy of this method by assessing the correlation between pre-treatment Tc-99m MAA activity distribution and post-treatment Y-90 activity distribution using relative SUV (standardized uptake value) thresholds.
Methods: 11 patients injected with Y-90 microspheres having both pre-treatment Tc-99m MAA and post Y-90 bremsstrahlung SPECT/CT scans were selected. Activity distributions were qualitatively assessed for agreement. Four contours were generated on each scan based on a relative percentage of the maximum SUV (20%, 40%, 60%, 80%). A rigid registration was done to transfer the post contours to the pre SPECT/CT scan. Finally, the Dice coefficient and Hausdorff distance were calculated between the pre and post threshold contour for each respective level.
Results: Qualitatively, the pre and post activity distributions visually agreed. The average Dice coefficient was 0.52 (range: 0.29-0.76), 0.27 (0.01-0.51), 0.15 (0-0.43) and 0.09 (0-0.65) for the 20%, 40%, 60% and 80% thresholds, respectively. The average Hausdorff distance was 1.14 cm (range: 0.69-1.46 cm), 1.88 cm (0.63-4.4 cm), 2.95 cm (0.41-10.56 cm) and 4.34 cm (0.44-11.21 cm) for the 20%, 40%, 60% and 80% thresholds, respectively. For the 80% threshold, only 2/11 cases had a Dice coefficient >0.25 demonstrating that the region of high SUV did not correlate well. The maximum Dice coefficient was 0.76 and the minimum Hausdorff distance was 0.69 cm, both using the 20% threshold.
Conclusion: There does not appear to be good correlation between SUV in the pre Tc-99m MAA scans and the post treatment Y-90 bremsstrahlung scans for any threshold value. Tc-99m MAA scans for Y-90 treatment planning should be used with caution.