During the reconstruction of radionuclide activity of waste by conventional tomography gamma scanning (TGS), the measurement time is long due to the division of a large number of invalid voxels, and the reconstruction accuracy is low due to the iterative solution of the pathology equation. The traditional division method of encrypting only circumferential voxels does not significantly improve the reconstruction accuracy.

This study aims to solve the problem of long measurement time and low accuracy of TGS by elaborating the influence law of voxel division method on reconstruction accuracy.

For a 400 L cement waste drum with a radius of 35 cm, a coaxial HPGe detector with a detection efficiency of 40% (crystal diameter 6.09 cm, length 5.18 cm) was used at a distance of 74 cm from the center of the drum to measure two types of nuclides, ⁶⁰Co and ¹³⁷Cs. The error situation of the optimized method of voxel partitioning and the traditional voxel partitioning method were compared by random point source activity reconstruction experiments. By calculating the condition number and count rate deviation of different voxel partitioning methods, the influence law of voxel partitioning methods on the reconstruction accuracy was investigated.

Comparison results show that the radial encryption method leads to an increase in the number of very small value points and a decrease in the value of very large value points in the count rate deviation curve, which improves the reconstruction accuracy. Compared with the conventional division method, the voxel division optimization method reduces the number of voxels and thus the measurement time by about 9/10; at the same time, it reduces the number of conditions and the count rate deviation, which reduces the maximum reconstruction error by about 2/3.

An optimized voxel division method proposed in this study makes use of the influence law of different voxel divisions on the reconstruction error through theory and experiment, hence improves the measurement accuracy and reduce the measurement time.