Voxel based internal dosimetry during radionuclide therapy

Many radionuclides have been used for several decades in cancer treatment. (131)I, (90)Y, (89)Sr, (111)In, (177)Lu and (223)Ra are some of the most widely used radioisotopes. Therapeutic results and side effects can be associated only if the absorbed dose is well estimated. Knowledge of the absorbed dose during radionuclide therapy is the only method that can compare therapeutic results between different therapeutic techniques such as external radiotherapy and radio immunotherapy. Accurate patient specific estimation of the absorbed dose to the tumor and normal tissue can be achieved with voxel based internal dosimetry.

The aim of this study is to develop a computer algorithm that calculates absorbed dose at every voxel of quantitative SPET scintigraphy image and establish a general internal dosimetry protocol for therapy by radionuclides.

SPET scintigraphy images of known (131)I activities were obtained. The known activity of (131)I was contained in a cylindrical phantom of 16cm radius. A numerical factor was determined to convert the measured count rate from the SPET images to activity. The algorithm that calculate absorbed dose at every voxel of the scintigraphy image was developed by MATLAB. MATLAB is a high level computer language with interactive environment and performs mathematical calculations by matrices. The scintigraphy images were imported in MATLAB and were converted to 3-dimensional matrices. Every element of the matrix was assigned with the respective count rate. The matrix was multiplied with the conversion factor and the new elements represented the activity at every voxel. A cumulative activity matrix was made from activity matrices that were obtained at different time points. The absorbed dose at every voxel of the cumulative activity matrix was computed with the convolution method. A3- dimensional convolution matrix with size 5x5x5 was created. The elements of this matrix are numerical factors that convert cumulative activity to absorbed dose. The convolution between these two matrices was made using the 'convn' function of MATLAB. This function convolves the cumulative activity matrix with the convolution matrix and gives a new matrix with numeric elements that represent absorbed dose at every voxel. Furthermore, regions of interest can be drawn on the images and dose volume histograms and mean absorbed doses can be calculated.

SPET scintigraphy images were obtained for 64MBq (131)I activity. A cumulative activity of 230GBq*s was calculated and the voxel convolution method showed that mean absorbed dose was 20.6Gy. Maximum dose of 50Gy was calculated for the central voxel in the activity. A region of interest was drawn and the dose volume histogram was calculated. The results were compared to MIRDOSE3.1 program. A mean dose of 20.1Gy was calculated with MIRDOSE3.1 for the same cumulative activity in 0.85cm diameter.

Patient specific voxel internal dosimetry can be performed accurately with MATLAB. Dose volume histograms and mean dose can be calculated.