Poster Presentation 36th TROG Cancer Research Annual Scientific Meeting 2024

Clinical evaluation of a real-time treatment verification system for stereotactic body radiotherapy (#106)

Jose A Baeza-Ortega 1 , Christopher Cassell 1 , Danny Lee 1 , Peter B Greer 1 2
  1. School of Information and Physical Sciences. College of Engineering, Science and Environment., University of Newcastle, Newcastle, NSW, Australia
  2. Radiation Oncology, Calvary Mater Newcastle Hospital, Newcastle, NSW, Australia

Background

Stereotactic body radiotherapy (SBRT) is characterised by highly hypofractionated schemes. The large doses per fraction in SBRT intensify the need for error detection during delivery. Real-time transit dose verification using electronic portal imaging device (EPID), is the only in-vivo dosimetry methodology able to detect errors during treatment delivery.


Aims

The aim of this work is to present the results of the clinical deployment of a real-time delivery verification system for SBRT.


Methods                                                 

The verification system uses a dedicated workstation equipped with a frame-grabber card to acquire a real-time stream of image frames from the Varian aS1200 EPID panel. EPID frames acquired during treatment delivery are compared to predicted control-point interval EPID images. Predicted images are obtained using a physics-based analytical prediction model for a 6XFFF beam, modified from a previously presented dual source fluence model. Real-time analysis of machine parameters is performed.

 

The verification system is largely automated requiring only a patient selection and beam-order confirmation from the therapist. 19 SBRT patients, including spine, pelvic, prostate, and lung, were included in the study.  

 

Results

Seventy-one fractions were successfully verified. Non-verified fractions were the result of therapist forgetting to set up the system, and technicalities during delivery produced by delivery interruptions and storage issues.

Qualitative retrospective analysis indicated that:

  • Average signal differences for 16 out of the 19 patients were within +/- 5%, and only one patient presented control points with signal differences above +/- 15%.
  • Signal differences were very consistent among treatment fractions.
  • No correlation in signal differences was found when categorising patients per treatment site.
  • Among lung patients, larger signal differences were associated with greater anatomical variations of tumours due to breathing motion.  

 

Conclusions

A software platform to analyse real-time delivery at a control-point level was developed, demonstrating sufficient performance to operate in the clinical practice.