ESTRO 2024 - Abstract Book

S4072

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2024

deformably propagated from pCT to each pre-fraction scan and, when necessary, edited manually by the same physician who contoured the pCT. The planned doses were re-calculated on the deformed anatomy using the fiducial markers alignment center to rigidly register the pCT to each pre-fraction scan (Figure 1). Constraints of the deformed OARs were considered met at V35Gy<1 cc. When 2 pre-fraction CT scans were acquired, average V35Gy values were considered. Data were analyzed to define a threshold for EIV values: EIV above or below threshold should predict when constraints are likely to be or not to be exceeded.

Results:

All planned dose distributions met the initially required duodenum, stomach and bowel constraints (V35Gy<0.5 cc). EIV values calculated on the planning CT anatomy ranged from 0 to 21.4 cc (median 1.2 cc). Comparison between observations on pre-fraction deformed anatomy and EIV-based predictions is shown in figure 2. The recalculated doses on repeated CT scans met constraints (V35Gy<1cc) for the three OARs in 7/20 patients. An optimal threshold EIV value of 1.5 cc was identified. When EIV values for duodenum, stomach and bowel calculated on the pCT were all below 1.5 cc, dose constraints (V35Gy<1 cc) were never violated for the 3 deformed OARs contoured on pre-fraction CT scans (Figure 2). On a patient basis, 3 false positive values were observed (that is EIV> 1.5 cc but no constraint violation observed) and EIV prediction rate was 85% . Considering each OAR separately, prediction rate using a 1.5 cc threshold was the highest for duodenum (80%) and lower for stomach and bowel (75%).