RT Journal Article SR Electronic T1 Collapsed-cone convolution superposition for improved accuracy of voxelwise dosimetry JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 535 OP 535 VO 61 IS supplement 1 A1 Stephen Graves A1 Ashok Tiwari A1 John Sunderland YR 2020 UL http://jnm.snmjournals.org/content/61/supplement_1/535.abstract AB 535Objectives: Convolution-based voxelwise dosimetry for radiopharmaceuticals has gained popularity in recent years, however this strategy does not account for tissue density and composition heterogeneities during energy deposition. This limitation is particularly impactful in the case of image-based bone marrow dosimetry. A refined convolution method - collapsed cone convolution superposition (CCCS) - has been used for decades in external beam radiotherapy dose calculation.[1] This technique has the potential to accurately handle heterogeneities without the substantial computational resources required for Monte Carlo-based dose calculation. The objective of this work was to implement the CCCS method for radiopharmaceutical dosimetry, and to compare the accuracy of this approach to that of convolution- and Monte Carlo-based dosimetry. Methods: A CCCS platform was developed in MATLAB. Required input data are as follows: quantitative SPECT volume, CT volume, HU to electron density calibration, and radioisotope of interest. Following dose calculation, an RT-Dose DICOM file is produced. To compute dose, energy is transported along a finite number of rays originating from the center of each voxel. Along each ray, tracing is performed to evaluate the effective pathlength through each subsequent voxel. Effective pathlengths are used to calculate the differential energy deposition along each step by interpolation within an integrated dose point kernel. Dose point kernels simulated in water, which we have we previously communicated, were used herein.[2] Scaling of effective pathlength was performed within voxels of density greater than 1 g/cm3 to account for increased photoelectric absorption in calcified (high-Z) tissues. For testing, quantitative SPECT/CT data was obtained following treatment of a patient with 177Lu-DOTATATE. Patient data was acquired under a protocol reviewed and approved by the University of Iowa Institutional Review Board. A dose rate map for this patient was computed using (1) our CCCS platform, (2) simple convolution, and (3) Monte Carlo simulation using MCNP v6.2. Results: When comparing against MCNP 6.2, CCCS dosimetry had improved accuracy over conventional convolution dosimetry in the bone marrow (-2.0% vs. -6.5%), spine (-4.0% vs -11%), and lungs (-14% vs. 61%). As expected, approximately equivalent results were observed in the kidneys (-0.5% vs.-0.5%), liver (-0.9% vs. -0.6%), and tumors (0.2% vs -1.1%). These results indicate that kernel stretching via the collapsed cone method is suitable for radiopharmaceutical dosimetry. More work is needed to confirm the utility of this technique in larger datasets and with other radioisotopes. Conclusions: We have developed a new dosimetry platform that offers increased accuracy in bone marrow and lungs. Bone marrow in particular is tissue of concern, and so this platform may facilitate the development of new image-guided therapy techniques. References: [1] Ahnesjö, A. "Collapsed cone convolution of radiant energy for photon dose calculation in heterogeneous media." Med. Phys., 16(4): 577-592, 1989. [2] Graves, SA, Flynn, RT, and Hyer, DE. "Dose point kernels for 2,174 radionuclides." Med. Phys., 46(11): 5284-5293, 2019.