ESTRO 2024 - Abstract Book

S5610

RTT - Patient care, preparation, immobilisation and IGRT verification protocols

ESTRO 2024

normalized to ensure that 98% of the planning target volume (PTV) received 98% of the prescribed dose (PTV_V98%=98%). Image-guided radiotherapy (IGRT) using cone beam CT (CBCT) was employed on each isocenter. Shifts and the overall quality of matching were evaluated by RTTs on a scale of 1-5. In the latest 18 patients, a second CT scan was performed four days before treatment (CT-4) to assess anatomical changes compared to CT-15 and evaluate their impact on the treatment plan and delivery (ClinicalTrials.gov identifier: NCT04976205). Non-rigid registration was used to propagate the PTV from CT-15 to CT-4, and the plan was recalculated using the same monitor units, tasks overseen by RTTs. The dose received by 95/98% of PTV (PTV_D95/98%) was calculated for both CT-15 and CT-4. Finally, the overall quality of matching was evaluated for CT-4 and compared to the CT-15. The statistical analysis was performed using the Wilcoxon Signed Rank test (p < 0.05 significant), with data collected and analyzed by an RTT as part of her master thesis.

Results:

A total of 775 CBCT scans were analyzed and ranked, of which the last 83 performed also the CT-4. No significant difference in the CBCT/CT-15 matching were observed between the first 692 and the last 83 scans, with median shifts of, respectively, 3.9[2.2-5.8] mm and 3.7[2.3-5.7] mm (p>0.05). Image quality agreement was sub-optimal (i.e., <4) in 13% of cases. The quality agreement of CBCT/CT-4 was higher with no cases ranked <4 (see figure 1).

Regarding the dose distribution, a significant reduction of PTV_D98% was observed (92.7% [90.7;95.0] for CT-4 vs. 98.0% for CT-15 - p<0.05) (see table 1). In two cases (i.e., 11% of cases), important differences were observed between the two CT scans, leading to plan re-optimization using CT-4 and subsequent treatment delivery. This rapid optimization was facilitated by automation tools utilizing artificial intelligence (AI) developed by our team