Analysis of diaphragm movements to specify geometric uncertainties of respiratory gating near end-exhalation for irradiation fields involving the liver dome

Background and Purpose The technique of gating near end-exhalation is commonly adopted to reduce respiration-associated geometric uncertainties for particle beam therapy. However, for irradiation fields involving the liver dome, how diaphragm movements generating liver–lung interface change, alongside geometric uncertainties, remain unspecified. Methods and Materials Patients receiving respiratory-gated computed tomography (RGCT) with four-dimensional computed tomography (4DCT) scans during simulation were retrospectively reviewed. Differences (Δ) between RGCT and 4DCT images, including diaphragm displacements and liver–lung interface changes, were investigated to specify geometric uncertainties during early inhalation phases. Craniocaudal displacements (Δy, in sagittal/coronal planes) of diaphragm segments (dorsal/ventral/right lateral/medial), liver area changes (ΔA, in axial planes), and liver extent changes in specific directions of incidence (Δr, in axial planes) were analyzed. Results Altogether, 162 patients received simulating RGCT and 4DCT scans. In 22 of them, both images involved the liver dome. For most cases during early inhalation phases, the Δy values in the dorsal diaphragm were significantly greater than those in the ventral diaphragm (p < 0.05), the ΔA values were significantly enlarged with inhalation progressing (p < 0.05), and the Δr values in the dorsal direction were significantly larger than those in the ventral direction (p < 0.05). These results suggested that the dorsal diaphragm moves earlier and more in a caudal direction than the ventral diaphragm during early inhalation phases. Conclusions For respiratory-gated radiotherapy near end-exhalation and irradiation fields involving the liver dome, components of geometric uncertainties are temporospatial, including diaphragm segment movements, inhalation phases of irradiation, and beam angles of incidence.