Influence of field-of-view and section thickness of diagnostic imaging on thermal neutron flux estimation in dose-planning for boron neutron capture therapy

Radiation treatment planning for boron neutron capture therapy (BNCT) often uses computed tomography (CT) images reconstructed utilizing various section thickness and field-of-view (FOV) settings. Based on these images, a geometrical model is created by setting material regions manually over the pixel space defined in the treatment planning system. Thus, a setting difference of several pixels inevitably occurs in creation of the model. The influence of different section thicknesses and FOVs on thermal neutron flux estimations using the BNCT planning system was studied here. A virtual phantom was created with six FOV sizes on the planning system. The position of the irradiated side of the phantom surface was shifted by 1-10 pixels along the beam direction or in the opposite direction to simulate the material setting on different pixels in the geometric model. The effect of a one-pixel-difference setting on thermal neutron flux increased with increasing FOV size. Next, a cylindrical and a spherical phantom were scanned, and each CT image was reconstructed with six FOV sizes and seven section thicknesses. The flux changes for all conditions were compared, with an allowable error rate of +/- 0.05, as in conventional X-ray radio therapy. The accuracy of neutron flux estimations was also evaluated by repeating the calculation procedures with CT scanning 5 or 10 times, and was found to be mostly within 0.03, except for the FOV-500 condition (0.074). These results suggested that a smaller FOV and section thickness with realistic conditions could improve evaluation accuracy of the thermal neutron flux for BNCT.