Assessment of the agreement between cone-beam CT and four-dimensional CT internal target volumes for stereotactic radiotherapy of lung cancer patients

Stereotactic radiotherapy is a novel approach to treat lung cancers with very large doses of radiation and high precision. Stereotactic lung radiotherapy is complicated by lung tumours moving during patient respiration. A motion compensation method for treatment planning is to use four-dimensional CT (4DCT) imaging to generate a maximum intensity projection (MIP) internal target volume (ITV). Image-guided radiotherapy during treatment may then involve acquiring a volumetric cone-beam CT (CBCT) image of the tumour and visually aligning it to the 4DCT MIP ITV contour. Imaging moving targets with CBCT results in blurry images and currently there are no studies reporting the effect that irregular breathing patterns have on CBCT volumes and their alignment to 4DCT MIP ITV contours. The objective of this project was to image a phantom moving with irregular breathing patterns to determine whether any configurations resulted in clinically significant errors in volume contouring or patient alignment.<br><br>This project used a Perspex thorax phantom to simulate a lung patient. Three wooden ‘lung’ inserts with Perspex ‘lesions’ were moved up to 4cm with computer-generated, and patient-specific breathing patterns. The phantom was imaged on 4DCT and CBCT with the same acquisition settings used for stereotactic patients in the clinic and volumes on all phantom images were contoured. This project assessed volumes for qualitative changes, and quantitative changes including volume, volume length, and alignment errors between CBCT volumes and 4DCT MIP ITV contours.<br><br>The results showed significant changes to 4DCT MIP and CBCT volumes for large amplitudes and irregular motion patterns. Artefacts presented in 4DCT MIP images, while CBCT volumes became significantly reduced in contrast. Both imaging modalities had reduced volumes up to 30% as well as shortened ITV lengths up to 3.7mm indicating that volume was underrepresented at the extremes of motion. When phantom lesions were moved with variable amplitudes from patient traces there was a distinct trend in reduced ITV with increasing amplitude that was not seen for constant amplitudes. In most cases sinusoidal motion patterns resulted in the closest agreements to known values and resulted in the smallest misalignments. Since most patients do not breathe sinusoidally this may lead to misinterpretation of previous studies using only sinusoidal motion.<br><br> Strategies are needed to compensate for volume underrepresentation at the extremes of motion for both 4DCT MIP and CBCT for larger and varied amplitudes, and for patterns with rest periods following exhalation. Lesions moving greater than 2cm had reduced 4DCT MIP ITVs and shortened ITV lengths that would warrant larger treatment margins added to the ITV as volume is underrepresented at the extremes of motion. Lesions moved with a rest period following exhalation were better aligned using an edge-to-edge alignment technique. Sinusoidal patterns gave the best agreements with known values, reinforcing the importance of investigating clinically relevant motions and their effects on 4DCT MIP and CBCT volumes. While the thesis confirms that using CBCT for stereotactic lung treatment verification is in most cases appropriate, limitations due to large and irregular motions were identified and methods for addressing these problems are suggested.<br>