Heartbeat simulation to evaluate intra-fraction dosimetric variations for 3 radiotherapy techniques
OC-0936
Abstract
Heartbeat simulation to evaluate intra-fraction dosimetric variations for 3 radiotherapy techniques
Authors: Nicola Maffei1, Bruno Meduri2, Francesca Itta3, Naveed Ashraf4, Maria Victoria Gutierrez1, Annalisa Bernabei1, Elena Zanella5, Francesca Di Pressa5, Elisa D'Angelo5, Ercole Mazzeo5, Frank Lohr5, Gabriele Guidi1
1Azienda Ospedaliero-Universitaria di Modena, Medical Physics, Modena, Italy; 2Azienda ospedaliero-Universitaria di Modena, Radiotherapy Unit, Dept. of Oncology, Modena, Italy; 3Azienda ospedaliero-Universitaria di Modena, Medical Physics, Modena, Italy; 4University of Modena and Reggio Emilia, Physics, Modena, Italy; 5Azienda Ospedaliero-Universitaria di Modena, Radiotherapy Unit, Dept. of Oncology, Modena, Italy
Show Affiliations
Hide Affiliations
Purpose or Objective
A heartbeat cycle model was developed to simulate the cardiac motion during radiotherapy (RT) and to assess the intra-fraction dosimetric effects on cardiac sub-structures comparing three planning techniques.
Material and Methods
A Hierarchical-Clustering atlas-based algorithm were developed, validated and later used to automatically segment 25 cardiac sub-structures in 10 breast CTs acquired in Breath Hold technique. The heartbeat cycle was simulated starting from ventricles’ volume variations data from literature. A commercially deformable image registration algorithm was used to drive the motion model obtaining 2 synthetic CTs at the maximum and minimum heart expansion during the cardiac cycle. All the cardiac sub-structures were automatically contoured in deformed image sets. Intra-fraction dosimetric variations were evaluated for a left breast target (266cGy x 16fx) within 3 RT techniques: 3D Conformal Radiation Therapy (3D-CRT), Volumetric Modulated Arc Therapy (VMAT), Helical Tomotherapy (HT) for a total of 90 plans calculated. Dosimetric parameters evauated for each Regions of Interest (ROIs) were the average and the maximum dose. Automation of the workflow was performed by integrating developed IronPython scripts into the treatment planning system for auto-contouring, creation of synthetic CTs and data extraction.
Results
The dosimetric analysis focused on ROIs closer to the high dose gradient in left breast target. 3D-CRT shows the lowest absolute Dose compared to the HT and VMAT but the wider range with the highest Dose values considering the total heart organ motion. In Left Ventricle (LV), VMAT shows the highest absolute dose compared to HT and 3D-CRT but relative doses are more robust. In Right Ventricle (RV), HT shows the highest absolute dose compared to VMAT and 3D-CRT with an improved target coverage and uniformity. In Proximal Left Anterior Descending Coronary Artery (ProxLADCA), Mid Left Anterior Descending Coronary Artery (MidLADCA) and Distal Left Anterior Descending Coronary Artery (DistLADCA), considering relative dose variations among cardiac cycle, 3D-CRT was the most affected by organ motion and VMAT was the most robust.
Figure 1 – Variations of average dose (up) and maximum dose (down) of the heart considering the maximum (orange) and minimum (blue) expansion during the simulated cardiac cycle. Comparison carried out for 3D, Tomotherapy and VMAT plans.
Conclusion
Automatic segmentation of cardiac sub-structures allows an accurate heart dose estimation in RT treatments. Deformable image registration algorithm can be used to simulate cardiac cycle. Robust optimization on synthetic image sets and organ motion models could increase accuracy in intra-fraction dosimetry of RT treatments.