Comparison of setup errors in three dimensions

The study analyzed 210 cone-beam computed tomography (CBCT) scans from 50 patients during radiotherapy. Displacement errors in the X, Y, and Z axes were recorded after each scan. Table 1 and Fig. 3 demonstrated that 18 (8.57%, 18/210) scans exhibited displacements exceeding 3mm in the X direction, 74 (35.23%, 74/210) in the Y direction, and 40 (19.05%, 40/210) in the Z direction. Table 2 summarized the setup error findings for the 210 CBCT scans across the 50 patients, indicating that the Y direction had the greatest error, followed by the Z direction, with the X direction exhibiting the least error. Applying the formula MPTV = 2.5∑ + 0.7σ, the expansion margins of the clinical target volume (CTV) in the X, Y, and Z directions were calculated to be 4.4mm, 6.4mm, and 5.8mm, respectively, as detailed in Table 3.

Table 1 Setup errors of 210 CBCT scans in 50 patients (mm)Fig. 3

3D scatter plot of set-up-error

Table 2 Absolute values of setup errors in 210 CBCT scans of 50 patients (mm)Table 3 Setup errors and MPTV analysis of 50 patients (mm)Effect of setup errors on dose distribution

Upon integration of setup errors into the TPS and subsequent dose recalculation, dosimetry to both the irradiated area and OARs exhibited substantial variability. Figures 4 and 5 depict the DVH of the IMRT dose distribution for a cervical cancer patient following setup error introduction. The original plan is denoted as Tr, with T1–T4 representing subsequent dose modifications due to setup errors (− 4.1, 4.0, 3.8) mm, (− 3.0, 3.8, 3.8) mm, (− 2.9, 2.7, 2.9) mm, and (− 2.9, 3.0, 1.8) mm, respectively. With setup errors of (− 4.1, 4.0, 3.8) mm, TC decreased from 95.2 to 88.4%, a 7.14% reduction. Setup errors primarily affected target coverage, with minimal impact on high-dose target regions; for OARs, the errors predominantly influenced the overall irradiation dose, as shown in Fig. 5.

Fig. 4

Original isodose line (Tr) and isodose line distribution after introducing errors (T1, T2, T3, T4)

Fig. 5

Original DVH (Tr) and DVH distribution after introducing errors (T1, T2, T3, T4)

The change of dosimetric parameters caused by setup errorsEffect of setup errors on dosimetric parameters of target volume

Table 4 displays the results of linear regression analysis examining the relationship between PTV dosimetry and setup errors in the X, Y, and Z directions. The analysis revealed that setup errors in the X direction did not significantly impact PTV dosimetry (P > 0.05). Conversely, setup errors in the Y and Z directions were found to significantly affect PTV_TC, PTV_HI, PTV_CI, and CTV_TC (P < 0.05). However, setup errors in the X, Y, and Z directions did not significantly influence CTV_HI and CTV_CI (P > 0.05).

Table 4 Linear regression analysis of PTV dosimetry and X/Y/Z direction error valuesCorrelation between setup errors and OAR dosimetry parameters

Table 5 reveals that linear regression analysis of intestinal dosimetry and setup errors in the X, Y, and Z directions showed no statistically significant impact of X-direction errors on Colon_Dmax, Colon_V30, Colon_V40, Colon_V50, Intestine_Dmean, Intestine_V50, and Rectum_Dmean (P > 0.05). However, X-direction errors did have a statistically significant effect on Intestine_Dmax, Intestine_V30, Intestine_V40, Rectum_Dmax, Rectum_V30, and Rectum_V40 (P < 0.05). For Y-direction errors, there was no significant effect on Intestine_Dmax, Intestine_V30, Intestine_V40, Rectum_Dmax, and Rectum_V40 (P > 0.05), but a significant effect on Colon_Dmax, Colon_V30, Colon_V40, Colon_V50, Intestine_Dmean, Intestine_V50, Rectum_V30, and Rectum_V50 (P < 0.05). Lastly, Z-direction errors did not significantly affect Colon_V40, Rectum_Dmean, and Rectum_V30 (P > 0.05), but did significantly impact Colon_Dmax, Colon_V30, Colon_V50, Intestine_Dmax, Intestine_Dmean, Intestine_V30, Intestine_V40, Intestine_V50, Rectum_Dmax, Rectum_V40, and Rectum_V50 (P < 0.05).

Table 5 Correlation analysis of intestinal dosimetry and X/Y/Z direction error values

Table 6 shows the results of linear regression analysis on femoral head and bladder dosimetry in relation to setup errors in the X, Y, and Z directions. The analysis found no significant effect of X-direction errors on FemoralHead_V30, Bladder_Dmean, Bladder_V30, and Bladder_V50 (P > 0.05), but did reveal significant effects on FemoralHead_Dmax, FemoralHead_V40, Bladder_Dmax, and Bladder_V40 (P < 0.05). Y-direction errors significantly affected Bladder_Dmean, Bladder_V30, and Bladder_V40 (P > 0.05), as well as FemoralHead_Dmax, FemoralHead_V30, FemoralHead_V40, Bladder_Dmax, and Bladder_V50 (P < 0.05). Z-direction errors did not significantly impact FemoralHead_V40, Bladder_Dmax, and Bladder_Dmean (P > 0.05), but did significantly affect FemoralHead_Dmax, FemoralHead_V30, Bladder_V30, Bladder_V40, and Bladder_V50 (P < 0.05).

Table 6 Correlation analysis of femoral head and bladder dosimetry and X/Y/Z direction error values