Optimising lung imaging for cancer radiation therapy

Effective radiotherapy is dependent on being able to (i) visualise the tumour clearly, and (ii) deliver the correct dose to the cancerous tissue, whilst sparing the healthy tissue as much as possible. In the presence of tumour motion, both of these tasks become increasingly difficult to perform accurately. This increases the likelihood of an incorrect dose being delivered to cancerous tissue and exposure of healthy tissue to unnecessary radiation. For tumours in the lung and thoracic region subject to respiratory-induced motion, 4D Cone-Beam CT (4D-CBCT) is a novel approach for producing a sequence of 3D images of the patient's anatomy throughout different phases of the respiratory cycle. However, current implementations involve sub-optimal heuristic approaches to acquire the imaging data required to account for tumour motion. This leads to undersampling of images for particular phases in the respiratory cycle (such as peak inhale and exhale), resulting in noisy or poorly reconstructed 3D images. In this paper we present a novel Mixed Integer Program (MIP) to optimise the timing and angles for the acquisition of imaging data. The result is greatly enhanced image quality for each image across the respiratory cycle, whilst minimising motion blur. Numerical experiments indicate that our approach universally improves over the conventional acquisition process by 93% and simultaneously reduces unnecessary dose to the patient and can be solved in under a minute. (C) 2019 Elsevier B.V. All rights reserved.