ESTRO 2023 - Abstract Book

S1316

Digital Posters

ESTRO 2023

PO-1621 An automated treatment planning method for prostate SABR

J. Wood 1 , A. McGowan 2 , W. Beasley 2 , M. Serra 3

1 The Christie NHS Foundation Trust, Christie Medical Physics & Engineering, Manchester , United Kingdom; 2 The Christie NHS Foundation Trust, Christie Medical Physics & Engineering, Manchester, United Kingdom; 3 The Christie NHS Foundation Trust, Clinical Oncology, Manchester, United Kingdom Purpose or Objective Prostate SABR treated in 5 fractions following the PACE trial is becoming increasingly popular compared to standard 20 fraction approaches. Acute toxicity and toxicity at 2 years has been shown to be comparable between SABR and conventional treatment. The reduction in treatment fractions offers many advantages including patient convenience, resource efficiency and promoting environmental sustainability. As radiotherapy centres adopt prostate SABR as standard of care, it is essential that high quality SABR treatment plans can be generated efficiently. The aim of this work, therefore, was to develop an automated treatment planning solution for prostate SABR that could easily be implemented into any treatment planning system. Materials and Methods Sixteen clinical prostate SABR treatment plans generated over a 6 months period were selected and used to create a model of OAR dosimetry. For this, OARs were subdivided based on their , shape and proximity/overlap with CTV_4000 and PTV_3625 and the differential DVHs of these subdivisions were averaged to produce a nominal differential DVH for each (see left hand side of Figure 1).

Figure 1: Flow diagram of prostate SABR auto-planning model training (left) and testing (right). Five new patients, that had not been included in the training cohort and for whom clinical treatment plans had been generated manually, were then selected as a test cohort. For each case in the test cohort, the OARs were subdivided (as per the training dataset) and the subdivision volumes were used to scale the nominal DVHs. These were combined and converted to cumulative DVHs using an in-house Java application. The resultant DVHs were used to customise the optimisation objectives of a standard class-solution in Pinnacle to drive the treatment plan optimisation of auto-plans (see right hand side of Figure 1). Results Four out of five of the automatically generated treatment plans met all of the PACE trial dose criteria, had acceptable target coverage and improved OAR sparing compared to corresponding manual-plans and a standard class-solution. In particular, rectum V18.1 Gy and V29 Gy and bladder V18.1 Gy were significantly lower in the auto-plans than in