ESTRO 2021 Abstract Book

S700

ESTRO 2021

between the patient model used for online dose calculations and the actual anatomy of the patient during delivery. We present cases to illustrate the clinical value of in vivo portal dosimetry as an independent end-to- end check. Materials and Methods The online patient model is obtained either by a rigid transformation of the reference planning CT to the online MR-image in the Adapt To Position (ATP) workflow, or by contour propagation with bulk electron density assignment in the Adapt To Shape (ATS) workflow. Between June 2020 and February 2021, 498 online adapted plans (326 prostate, 68 rectum, 69 liver, 35 lymph node oligo metastases corresponding to 56, 11, 25, 7 patients respectively) were verified by comparing 3D EPID-reconstructed and planned dose distributions using g-analysis (3% global/3mm/50% threshold) and the dose difference value at a dose reference point (DRP). Tolerance limits of 85%, 0.7 and 4% were used for the γ passrate, mean γ and ΔD DRP , respectively. Results 76% of plans showed all parameters within tolerance. For the majority of alerted cases, the detected deviations were due to limitations of the method, e.g. sites with large inhomogeneities, or a suboptimal location of the DRP. In some cases, however, the detected deviations were real, three examples of which follow. 1. A single adaptation of a prostate treatment presented a γ-pass of 58.7%, far below the average treatment value of 98.1%. Investigation revealed that the anterior MRI receive coil was accidentally pushed towards the feet of the patient during treatment. As a result, several beams were attenuated by the supporting arm of the coil before reaching the patient, see Figure 1. 2. Poor results were reported for an ATP adaptation of a prostate treatment (γ-pass of 75.2% versus a treatment average of 97.2%). Figure 2a shows the registration between the reference CT and the online MRI, suggesting that this adaptation would be a candidate for the ATS workflow. 3. In a rectum treatment (ATS workflow), air pockets were not considered for online dose calculation as they were not present in the reference CT and therefore not warped, see Figure 2b. The dosimetric effect of the ‘missing’ air volume (over-dosage) was detected by EPID dosimetry.

Conclusion EPID dosimetry can detect both patient related deviations and workflow incidents that other plan QA methods would have missed. EPID dosimetry acts as an independent end-to-end check and provides the MR-guided adaptive workflow with an extra safety net.

PD-0865 The delta index: a novel metric to assess dose accumulation uncertainty in MR-guided