OLINDA based dose calculation using planar and SPECT images versus Monte Carlo as the gold standard

Objectives As interest in the therapeutic use of radiopharmaceuticals grows, the demand increases for accurate dosimetry protocols. Our objective was to investigate the accuracy of different dose estimations.

Methods Two 8 voxel volumes of I-131 were digitally inserted inside patient CT data near the spine and in the lung to represent activity in tumours. The Monte Carlo EGSnrc user-code, DOSXYZnrc, was used to accurately calculate the dose to the tumours and surrounding normal tissue from the inserted activity distribution. Planar study (anterior and posterior) and SPECT data were also generated. The cumulated activity was estimated from the planar and SPECT images with and without attenuation correction (AC) and used as input for OLINDA. Tumour masses were determined using CT images and compared to those derived from NM images. OLINDA doses were compared to the Monte Carlo calculations considered as the gold standard.

Results OLINDA dose estimations depend strongly on the accuracy of activity and tumour mass determination. For true tumour volumes (masses), the planar study errors range from 400% without AC to 5-10% errors with AC. For SPECT, the errors range from 300% without AC to 20% with AC. When tumour volumes were obtained from NM images the errors increased by up to a factor of 10. Only EGSnrc was able to provide dose estimates to the surrounding tissue.

Conclusions Both planar and SPECT based OLINDA dose calculations using exact tumour masses (CT-based) and quantitative activity provide doses within 10-20% of true values, while lack of AC and incorrect mass estimation result in dramatic inaccuracies.