Optimisation of image guidance for improving radiation delivery in children's cancer

  • Abigail Bryce-Atkinson

Student thesis: Phd


Cone beam CT (CBCT) is the most common technique used for modern image guided radiotherapy (IGRT). Children are more radiosensitive and at greater risk of radiation-induced malignancies than adults. Adult CBCT exposures can lead to excessively high doses in children and may not yield optimal image quality due to their differences in size and anatomy. Currently, protocols tailored to children are lacking, and the limits of dose reduction have not yet been studied. This thesis aims to evaluate the feasibility of introducing "ultra-low" dose CBCT protocols, such that children can benefit from highly accurate treatment delivery, without concern over additional imaging dose. In this thesis, a method was developed to simulate low dose CBCT using clinical ("standard dose") scans as input. This allowed for new protocols to be assessed in patients without additional "experimental" scans, and hence no additional dose. Image quality was assessed qualitatively and quantitatively throughout this thesis. Image quality assessment directly relevant to the IGRT task was performed by visual grading analysis (VGA) and by comparing registration accuracy to current protocols. To further explore the limits of dose reduction, a novel method of image quality assessment was used in acquired and simulated paediatric CBCT scans (ranging from ultra-low to adult doses) to determine the noise contributions of quantum noise (from dose reduction), and "anatomical noise" arising due to motion (e.g. respiratory or bowel motion) and artefacts (e.g. gas in the abdomen). Novel hardware was developed, since the lowest CBCT exposures cannot yet be clinically implemented. The paediatric bowtie filter was designed with built-in attenuation and the shape tailored to children to achieve ultra-low doses. The impact of the filter use was evaluated in paediatric patients by further developing the simulation method. A combination of simulated and acquired scans were used to test the limits of reduced scan time (and dose) 4D CBCT, assessed in lung cancer patients. Results showed that paediatric CBCT dose can be reduced down to 16% (0.125mGy) of the current paediatric protocol (0.8mGy), while maintaining accurate visual assessment and registration of bony anatomy. The novel filter resulted in five-fold dose reduction compared to the standard "adult" filter, and image quality remained appropriate for setup on bony anatomy even at the lowest doses. In contrast, soft tissue registration was accurate for moderately reduced doses, performing worse for doses below 1mGy. Increasing paediatric CBCT dose above 1mGy held no benefit in improving image quality, since above this threshold, anatomical noise is the dominating component of total noise. 4D CBCT scan time could be halved whilst maintaining accurate 4D registration to the tumour. Further reduction caused considerable degradation in visual quality and performance of 4D registration. Overall, the results in this thesis have informed recommendations for ultra-low dose CBCT protocols for children that can be easily implemented in clinic by choosing appropriate exposure settings and/or use of the novel paediatric bowtie filter, whilst maintaining image quality suitable for IGRT purposes. Further, the short scan time 4D CBCT protocol was implemented clinically at The Christie NHS Foundation Trust.
Date of Award31 Aug 2021
Original languageEnglish
Awarding Institution
  • The University of Manchester
SupervisorGillian Whitfield (Supervisor) & Marcel Van Herk (Supervisor)


  • image guided radiotherapy
  • cone beam CT
  • radiotherapy
  • paediatric

Cite this