Enhancing thoracic imaging for lung cancer radiation therapy: translating adaptive imaging and reconstruction to the clinic

Abstract

Radiation therapy or radiotherapy is a highly effective and economical method for treating cancer. Safe and effective radiotherapy is facilitated through the precise delivery of radiation to the tumour whilst minimising radiation exposure to healthy tissue. Radiation therapy relies on image guidance technologies to visualise tumour movement prior to treatment delivery. Currently 4-Dimensional Cone Beam Computed Tomography (4DCBCT) imaging is utilised to gain knowledge of how patient anatomy will move during treatment. However thoracic tumours are affected by respiratory motion which can corrupt image quality. This thesis deals with a new approach to acquire 4DCBCT images where the imaging system adjusts in real-time according to patient respiratory signals to maintain image quality and reduce imaging radiation dose. This approach is called adaptive 4DCBCT.

This thesis encompasses four studies. The first study presents clinical trial results of the first 10 patients of the adaptive 4DCBCT clinical trial (ACTRN12618001440213), evaluating and establishing image reconstruction methods that would best complement the proposed adaptive acquisition method.

The second study assesses the viability of reducing the imaging dose and time for adaptive 4DCBCT without impairing image quality.

The third study evaluates how gantry velocity and angular separation impact 4DCBCT image quality on ultrafast imaging systems like the Varian Halcyon.

The final study proposes a machine learning motion-compensated reconstruction framework to accelerate the image reconstruction process for adaptive 4DCBCT to aid its translation to the clinic.

The body of work demonstrates the advancement and capabilities of adaptive 4DCBCT imaging, enabling image quality improvements for next generation image guided radiotherapy. The work here paves the way for future commercial clinical implementation of these technologies.