Session Item

The goal of this study was to assess breast volume changes by investigating dose deviations and their influence on complication risk assessment in breast cancer patients undergoing photon radiotherapy employing CBCT based synthetic CTs generated with four different methods.

Seventy-six breast cancer patients treated with a partial VMAT photon technique (70% conformal, 30% VMAT) were included in this study. Forty-three patients received a fractionation of 21x2.17Gy(+0.49Gy) while the other thirty-three received 16x2.66Gy. All patients showed more than 5mm swelling or shrinkage of the breast on the CBCT compared to the planning-CT (pCT), followed by a repeat-CT (rCT) for dose evaluation. The original CBCT was corrected using four different methods: 1) HU-override correction (orgCBCT), 2) conversion correction (cCBCT), 3) deep learning (DL) correction (sCT) and 4) virtual correction (vCT). Thirty-seven patients were used for training and validation of the DL model. The remaining thirty-nine patients have been used for dose calculation and HU accuracy analysis on the pCT, rCT and the four created synthetic CTs. HU-accuracy analyses consisted of calculating the mean absolute error (MAE) and average (AVG) within the whole breast CTV and the field of view of the CBCT minus 2 cm (CBCT-ROI), rigidly copied from the rCT to ensure equal dimensions over all image sets. Contours of the Clinical Target Volume (CTV) and heart from the pCT were warped to all image sets and the dose was recalculated to determine relevant dose-volume parameters: D1, D2, D95, D98, D99 and mean dose. The mean heart dose was used in combination with the normal tissue complication probabilities model for acute coronary events to assess clinical impact of the dose variations between the four sets of synthetic CTs.

Figure 1 shows images for two different patients, together with the calculated dose distribution and the regions of interest. The HU-accuracy analyses showed larger HU deviations for the CBCT-ROI than for the whole breast CTV, on average an increase of 46% was observed due to the inclusion of more heterogeneous tissues. Figure 2 shows the dose comparison for all dose statistics for the whole breast CTV. The dose difference as percentage of the prescribed dose between the rCT and pCT is on average -0.3% (±0.8%), which is considered to be the golden standard. Only the orgCBCT-pCT deviates significantly (p<0.001) over all dose statistics compared to the golden standard, resulting in 1.6%(±0.9%) average dose deviation. The mean heart dose is similar for all image sets and varies at max 7 cGy, which corresponds to a min-max change in risk of an ACE by 0.01%-0.06%.

The conversion correction, deep learning correction and virtual correction can be applied for an accurate synthetic CBCT dose evaluation for photon radiotherapy breast cancer patients as there is no statistical difference relative to the golden standard. Future work is warranted for breast proton therapy.