Session Item

To develop and evaluate a fast, fully automated multi-criteria treatment planning strategy for adaptive HDR-BT for locally advanced cervical cancer. This automated strategy avoids suboptimal and slow manual treatment planning.

Our in-house developed TPS for automated multi-criteria treatment planning was extended with an option for combined intracavitary + interstitial (IC+IS) cervical cancer BT, and connected to the clinical treatment planning system for plan evaluation. The algorithm was configured using 22 single-fraction (SF) IC+IS training plans. Special attention was paid to establishing the clinically desired ‘pear-shaped’ dose distribution. Autoplanning was evaluated on 63 other SF IC+IS cases by blind clinician comparisons with corresponding clinical plans (SFclin). Subsequently, we developed an adaptive scheme for automatic planning of all IC+IS BT fractions of a patient, considering dose delivered in previous EBRT and BT fractions. The effect of adaptive autoplanning on total treatment (TT) plans (external beam + 3 BT fractions) was evaluated for 16 patients by simulating the clinically applied adaptive strategy to generate TTauto plans and compare them with the corresponding clinical treatments (TTclin).

All automated SF (SFauto) IC+IS plans were clinically acceptable. The clinician''s plan comparisons pointed strongly at an overall preference for the automated plans: in 60/63 cases SFauto was preferred over SFclin, in 2/63 cases the overall quality of the plans was considered equal, and for 1/63 cases SFclin was preferred over SFauto. When comparing total adaptive treatments, the mean D90 CTVHR for the TTauto plans improved by + 3.6 Gy, (range +1.4 Gy - +6.0 Gy, p<0.005). The dose in the bladder and rectum was significantly reduced in the TTauto plans compared to TTclin with a mean improvement in D2cc of -0.9 Gy (range -6.6 Gy - +1.8 Gy, p=0.05) and -1.4 Gy (range -6.3 Gy – +4.9 Gy, p=0.04), respectively. There were no significant differences in D2cc of the sigmoid and small bowel (p=0.3). The average optimization time for autoplanning was 19.7 seconds (range 4.4 - 106.0 s).

Fast automated multi-criteria treatment planning for adaptive IC+IS HDR-BT for patients with locally advanced cervical cancer is feasible. High-quality treatment plans could be automatically generated within a clinically acceptable time frame (~20 sec). The observed improvement in dosimetric parameters, mainly the improvement of the dose to the CTVHR, is clinically relevant. The algorithm will be extended with an approach for optimization of the needle configuration, which could allow real-time interactive intra-operative treatment planning.