Optimization by visualization of indices

Physical 3D treatment planning provides a pool of parameters describing dose distributions. It is often useful to define conformal indices to enable quicker evaluation. However, the application of individual indices is controversial and not always effective. The aim of this study was to design a quick check of dose distributions based on several indices detecting underdosages within planning target volumes (PTVs) and overdosages in normal tissue.

Dose distributions of 215 cancer patients were considered. Treatment modalities used were three-dimensional conformal radiotherapy (3DCRT), radiosurgery, intensity-modulated radiotherapy (IMRT), intensity-modulated arc therapy (IMAT) and tomotherapy. The volumes recommended in ICRU 50 and 83 were used for planning and six conformation and homogeneity indices were selected: CI, CN, C_ICRU, COV, C_∆, and HI. These were based on the PTV, the partial volume covered by the prescribed isodose (PI; PTV_PI), the treated volume (TV_PI), near maximum D₂ and near minimum D₉₈. Results were presented as a hexagon—the corners of which represent the values of the indices—and a modified test function F (Rosenbrock’s function) was calculated. Results refer to clinical examples and mean values, in order to allow evaluation of the power of F and hexagon-based decision support procedures in detail and in general.

IMAT and tomotherapy showed the best values for the indices and the lowest standard deviation followed by static IMRT. DCRT and radiosurgery (e.g. CN: IMAT 0.85 ± 0.06; tomotherapy 0.84 ± 0.06; IMRT 0.83 ± 0.07; 3DCRT 0.65 ± 0.08; radiosurgery 0.64 ± 0.11). In extreme situations, not all indices reflected the situation correctly. Over- and underdosing of PTV and normal tissue could be qualitatively assessed from the distortion of the hexagon in graphic analysis. Tomotherapy, IMRT, IMAT, 3DCRT and radiosurgery showed increasingly distorted hexagons, the type of distortion indicating exposure of normal tissue volumes. The calculated F values correlated with these observations.

An evaluation of dose distributions cannot be based on a single conformal index. A solution could be the use of several indices presented as a hexagonal graphic and/or as a test function.