TY - JOUR T1 - High Performance Virtual CT for Enhanced Targeted Radionuclide Therapy Dosimetry JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 1303 LP - 1303 VL - 58 IS - supplement 1 AU - Greta Mok AU - Tiantian Li Y1 - 2017/05/01 UR - http://jnm.snmjournals.org/content/58/supplement_1/1303.abstract N2 - 1303Objectives: Previously we showed that non-rigid registration on sequential CT images is superior to sequential SPECT in 3D targeted radionuclide therapy (TRT) dosimetry (1), and proposed a virtual CT (vCT) method for improved dose estimation with only 1 CT acquisition (2). This study aims to (i) optimize the vCT generation protocol, and (ii) evaluates its performance as compared to sequential real CT (rCT) and conventional single CT (1CT) protocol in TRT dosimetry.Methods: We modeled a population of 9 phantoms with various anatomies and In-111 Zevalin activity distributions using the digital 4D XCAT phantom with random organ and whole-body movement between scans. We simulated SPECT acquisitions at 1, 12, 24, 72 and 144 hrs post-injection using an analytical MEGP projector, modelling attenuation, scatter, and collimator-detector response. Noisy projections were reconstructed using OS-EM algorithm incorporating full compensations. The corresponding attenuation maps of the phantoms served as the rCTs. To determine the optimal time point for the 1 CT acquisition, we generated vCTs at other time points by registering (i) CT to SPECT; (ii) SPECT to CT and (iii) SPECT to SPECT whole body images using the affine+b-spline method, using images at different time points as reference. Difference images and average normalized mean square errors (NMSE) were calculated between different vCTs and the corresponding rCTs. Organ-by-organ non-rigid registrations for rCTs and optimized vCTs were performed and the resultant motion vectors were later applied to register the corresponding SPECT images. For 1CT protocol, sequential whole body SPECT registrations were performed while using the single CT for organs segmentation. Voxel-by-voxel integration was applied on registered images over 5 time points, followed by convolution with a Y-90 dose kernel to generate dose images. Organ dose error (%ODE) and normalized absolute error (%NAE) of the differential dose volume histograms (DDVH) were measured for liver, spleen, kidneys and lungs.Results: For vCT generation, SPECT to SPECT registration shows the lowest NMSE using the 1st time point SPECT image as the reference. For %ODE, the results of vCT were similar to those of rCT registrations and were superior to 1CT, i.e., -0.24±1.56% vs. -0.49±1.76% vs. -6.37±5.63% for liver, -8.07±5.87% vs. -5.59±5.25% vs. -8.61±6.93% for spleen, -0.69±2.74% vs. -1.05±2.89% vs. -4.87±4.35% for kidneys, and -0.73±5.15% vs. -1.93±1.94% vs. 28.46±6.99% for lungs respectively. Results of %NAE were consistent with %ODE.Conclusion: The errors of vCT were minimized as compared to rCT for SPECT to SPECT registration when SPECT image with superior image quality, i.e., 1st time point image in this study, was used as the reference image. The use of vCT can effectively improve the TRT dosimetric results and are comparable to rCT, with potentials to reduce the substantial increase of radiation burden from repeated CT scans. Research Support: Fundo para o Desenvolvimento das Ciencias e da Tecnologia, Macau (079/2011/A3) Universidade de Macau (MYRG077(Y1-L2)-FST12-MSP) Universidade de Macau (MYRG2016-00091-FST) National Natural Science Foundation of China (81601525) ER -