Assessment of CBCT–based synthetic CT generation accuracy for adaptive radiotherapy planning

Abstract

Purpose Cone‐beam CT (CBCT)–based synthetic CT (sCT) dose calculation has the potential to make the adaptive radiotherapy (ART) pathway more efficient while removing subjectivity. This study assessed four sCT generation methods using 15 head‐and‐neck rescanned ART patients. Each patient's planning CT (pCT), rescan CT (rCT), and CBCT post‐rCT was acquired with the CBCT deformably registered to the rCT (dCBCT). Methods The four methods investigated were as follows: method 1—deformably registering the pCT to the dCBCT. Method 2—assigning six mass density values to the dCBCT. Method 3—iteratively removing artifacts and correcting the dCBCT Hounsfield units (HU). Method 4—using a cycle general adversarial network machine learning model (trained with 45 paired pCT and CBCT). Treatment plans were created on the rCT and recalculated on each sCT. Planning target volume (PTV) and organ‐at‐risk (OAR) structures were contoured by clinicians on the rCT (high‐dose PTV, low‐dose PTV, spinal canal, larynx, brainstem, and parotids) to allow the assessment of dose–volume histogram statistics at clinically relevant points. Results The HU mean absolute error (MAE) and minimum dose gamma index pass rate (2%/2 mm) were calculated, and the generation time was measured for 15 patients using the rCT as the comparator. For methods 1–4 the MAE, gamma index analysis, and generation time were as follows: 59.7 HU, 100.0%, and 143 s; 164.2 HU, 95.2%, and 232 s; 75.7 HU, 99.9%, and 153 s; and 79.4 HU, 99.8%, and 112 s, respectively. Dose differences for PTVs and OARs were all <0.3 Gy except for method 2 (<0.5 Gy). Conclusion All methods were considered clinically viable. The machine learning method was found to be most suitable for clinical implementation due to its high dosimetric accuracy and short generation time. Further investigation is required for larger anatomical changes between the CBCT and pCT and for other anatomical sites.

Simulation and pre-planning omitted radiotherapy (SPORT): a feasibility study for prostate cancer

“…However, the mean absolute difference in HU between the sCT and sCBCT/simCT of each test patient (as a measure of the quality of the sCT) were small with a median value of 74.7 HU (range 69.5 to 91.5 HU). This result aligns with a recent study that evaluated the quality of sCT among various generation methods [41]. On average, recalculating the on-couch plan on the sCBCT yielded a 0.7% reduction in PTV coverage and a 0.6% increase in the hotspot, and the dose differences of the OARs were minimal.…”

“…Thus, it is important to investigate the quality of the sCT and its effect on dosimetry. Previous studies evaluated the quality of sCT and the dosimeric accuracy for conventional CBCT based ART [40][41][42][43][44]. In this study, to achieve a similar goal, the test patients' simCT were used in the emulation studies as online CBCT data (sCBCT/simCT).…”

A hybrid method of correcting CBCT for proton range estimation with deep learning and deformable image registration

“…Accordingly, the individual CycleGAN sCT outperformed DpCT in MAE for some cases with relatively large internal gas volumes (patients 2, 5, 9, and 12 with 64.2-620.9 cc of internal gas; see tables 2 and 3). These results partly explain the previously reported mixed performances of deep learning-based sCT compared to that of DpCT (Thummerer et al 2020, O'Hara et al 2022 and support our region-selective use of the full DpCT (i.e. step 3) beyond the incorporation of only low-frequency components of DpCT.…”

Improving hybrid image and structure-based deformable image registration for large internal deformations

“…This is, for instance, used in adaptive radiotherapy (ART) for monitoring interfraction tumor motion, allowing for an accurate consideration of changes in the size and position of organs at risk (OARs) and gross tumor volumes (GTVs) (White et al 2019). DIR is also used in ART for mapping image densities from planning CTs to daily cone beam CTs (CBCTs) or MRs (Mannerberg et al 2020, O'Hara et al 2022. This enables treatment planning on treatment images that do not present electron density information.…”