Contouring
In order to simplify the experiments, the C-shaped target and Core (OAR) contoured on the water phantoms were regarded as the reference contours. The shape and relative position of the two structures were similar to those of spinal bone metastasis, which can be regarded as a preliminary study of clinical cases to a certain extent (Fig. 1), according to the TG-119 report [18]. The structures of the C-shaped target and Core were exported from treatment planning system (TPS) Raystation (Raysearch, Stockholm, Sweden) in the form of a DICOM file, and the position information of the contours were read by an in-house developed Python software; thereafter, the geometric transformations were carried out. Finally, the transformed structures were imported back to Raystation system in the form of a DICOM file. In order to analyze the contouring errors in detail, translation, scaling, rotation (except for the Core), and sine function transformation were simulated for both the C-shaped target and Core.
In this study, the translation transformations were divided into the following three cases: right, anterior, and posterior direction. Based on the location of the reference contour, at intervals of 1 mm, the data were moved 10 times to each of the right, anterior, and posterior directions to obtain the test contours (see Additional file 1: Figure A, B and C for the contours after the right, anterior, and posterior directions translation, respectively). Scaling transformation represents an equidistant expansion or reduction transformation in reference to the position of the contour. Considering the fast speed of scaling transformation changes, in the patient modeling module of the Raystation planning system, 10 equidistant transformations were performed at 0.5 mm intervals, excluding the anterior and posterior directions (see Additional file 1: Figure D and E for the contours after the expansion and reduction transformation, respectively). The rotation transformation involved taking the origin of the CT image coordinates as the rotation center point, using 1° as the interval, and rotating clockwise 10 times (see Additional file 1: Figure F for the contours after the rotation transformation). For the sine function transformation, we extracted the coordinate values (x0, y0) of the reference contour first, and then, we used the function y = sinωy0 (ω = 3, 4, 5, 6, …, 12) to carry out periodic transformations 10 times with a fixed amplitude (see Additional file 1: Figure G for the contours after the sine function transformation).
Geometric indices
In this study, we chose five widely used geometric indices for the evaluations, including three distance-type indices HD (maximum, mean, 95%) and two volumetric indices (DSC and Jaccard). These five geometric indices were calculated by 3DSlicer version 4.10.2 [19], which is open source software. The calculation of HD was performed on the RT-DICOM structures. The HD indices calculated by the 3DSlicer represent bi-directional distances, and the bi-directional distance is symmetrical; this type of distance is more stable than the unidirectional distance calculated by other methods.
Dosimetric indices
In this study, PTV and Core were regarded as the reference contours. Similarly, the IMRT plan, which meets the requirements for a simple version in the TG-119 report, was taken as the reference plan. The dose of 5000 cGy received by 90% of the target volume was taken as the prescription, and the dose grid was 2 mm. In order to determine the differences in dosimetric indices caused by different contour errors, the method adopted in this study was to use the existing dose distribution on the reference contour and overlay it on the geometrically transformed contour [20]. After geometric transformation, RTstructure was imported into the radiotherapy plan of the reference contour, and then, on the dose distribution, D98%, Dmean, and D2% of the PTV and Dmean and D2% of the Core were obtained. According to the ICRU-83 report [21], these dosimetric indices represent the minimum dose, mean dose, and maximum dose received by the target, and the mean dose and maximum dose received by the organs at risk, respectively. In this study, the dose differences ( ) of three dosimetric indices D98%, Dmean and D2% were calculated and normalized according to their respective clinical goals. Here, , where x represents the type of dosimetric index.
Analysis
An SPSS version 21.0 software (SPSS, Chicago, IL, USA) was used for linear regression analysis. The correlation coefficient R² was used to quantify the correlations between the geometric indices HD (maximum, mean, 95%), DSC, and Jaccard, and the dosimetric indices D98%, Dmean, and D2%. Two-sided P-values were obtained, and P-values <0.05 were considered significant. In addition, the geometric indices obtained from the right, anterior, and posterior directions of the PTV and Core translation transformations were compared, and the difference between them was tested for statistical significance using the Wilcoxon signed ranks test in SPSS, and from scatterplots of geometric indices versus dose difference, the feasibility of assessing the accuracy of test contours with geometric indices was analyzed.