Development of a comprehensive ‘automated’ workflow for radiotherapy treatment planning of spinal metastases using auto-segmentation and knowledge-based planning solutions


Principal research question: Does knowledge-based radiotherapy treatment planning provide a workable treatment planning solution for palliative treatment of spinal metastases?

VMAT is routinely used to treat radical radiotherapy patients at Guy’s Cancer. The use of this advanced and highly technical method in clinical practice is currently limited to radical patients due to the labour intensive nature of VMAT treatment planning. Palliative radiotherapy treatment accounts for approximately 40% of the radiotherapy workload at Guy’s Cancer and as such it is not possible to produce VMAT plans for this cohort of patients. Palliative treatment should be delivered in a timely manner to achieve durable symptom control with little toxicity and minimal patient inconvenience. As such the current standard of practice at Guy’s Cancer is to use conventional treatment planning, simple, single or parallel opposed beams of radiation to deliver a high dose of radiation to the target area. Sparing of OAR is achieved through basic multi-leaf collimation. The practicability of using VMAT rather than conventional planning for this large cohort of patients is limited due to the complexity and labour intensive nature of VMAT treatment planning, delivery and quality assurance. Knowledge-based planning has the potential to provide a comprehensive and streamlined treatment planning solution for this cohort of patients. Knowledge-based planning may provide improved consistency and efficiency in the treatment planning process and improved quality of treatment for palliative radiotherapy patients at Guy’s Cancer.


  • To determine the feasibility of atlas-based segmentation of the vertebral spine to aid delineation of the spinal metastasis target volume.
  • To validate atlas-based segmentation for delineation of OAR in the vicinity of the thoracic, lumbar and cervical regions of the spine.
  • To determine heterogeneity across spinal metastasis target volumes and dose to adjacent OAR for treatment plans produced using conventional radiotherapy approaches (single applied posterior, or two opposing anterior-posterior, open, un-modulated, radiation fields directed at the affected vertebral body).
  • To develop a DVH-estimation model for highly conformal radiotherapy treatment of a spinal metastasis to achieve consistent, high-quality treatment plans with superior coverage of, and homogeneity across, the target volume, and to determine dose to adjacent OAR.
  • To determine whether the aimed for ‘automated’ workflow is achievable in a timely manner.