Author: Jenny Dankelman, Ben J. M. Heijmen, Inger-Karine K. Kolkman-Deurloo, Remi A. Nout, Linda Rossi, Robin Straathof, Linda Wauben, Henrike Westerveld, Nick J. van De Berg, Sharline M. van Vliet - Perez 👨🔬
Affiliation: Department of BioMechanical Engineering, Delft University of Technology, Department of Gynecological Oncology, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Department of Radiotherapy, Erasmus MC Cancer Institute, University Medical Center Rotterdam 🌍
Purpose:
To quantify the dosimetric advantages of the 3D-printed patient-tailored ARCHITECT applicator with optimized needle channel configurations compared to clinically used intracavitary/interstitial applicators.
Methods:
First fraction treatment plans, MR images and delineations of 20 locally advanced cervical cancer patients previously treated with an intracavitary/interstitial applicator were acquired. The outer geometry of the patient-tailored ARCHITECT applicator was based on the patient's vaginal contour with a standardized needle entry section. Potential needle channels were automatically determined by optimizing Bézier curves between a given entry point and sampled interstitial segments in the target region. Using geometric coverage optimization, series of optimal configurations were automatically determined, with varying number of needle channels. Automated treatment plans were generated following EMBRACE-II planning aims with the in-house developed BiCycle software for the clinical (TPClin) and three ARCHITECT configurations: TPARCH,A = highest target coverage, TPARCH,B = highest coverage with same number of needles as clinical, and TPARCH,C = balancing target coverage and organ-at-risk (OAR) sparing.
Results:
The ARCHITECT applicators were generated in acceptable time (42±4 minutes). The median (IQR) number of needles used was 5 (4-6) in TPClin and TPARCH,B, 10 (9-10) in TPARCH,A, and 7 (6-8) in TPARCH,C. Treatment plans for the ARCHITECT configurations more frequently met CTV-THR D90 and D98, CTV-TIR D98, and OAR D2cm3 soft planning aims than for clinically used configurations (TPClin: 74% versus TPARCH,A: 92%, TPARCH,B: 83%, and TPARCH,C: 85% of instances), and were associated with dosimetric fraction improvements of >1.0 Gy in 9/16, 6/20, and 9/18 plans for TPARCH,A, TPARCH,B, and TPARCH,C respectively (4 TPARCH,A and 2 TPARCH,C plans did not converge).
Conclusion:
We demonstrated the dosimetric benefits of the 3D-printed patient-tailored ARCHITECT applicator. Our largely automated approach enables exploration of different trade-offs in the number of needles and dose plan quality.