ESTRO 2024 - Abstract Book

S4827

Physics - Quality assurance and auditing

ESTRO 2024

Figure 1. Clinical scores of H&N images from different centres

The phantom metrics which best correlated with high clinical image scores were, for H&N: high TTF50 (r=0.73, p = 0.003), contrast (r=0.58, p=0.003) and edge visibility scores of 1% target (r=0.70, p = 0.004); for prostate high TTF50 (r=0.83, p=0.002), low noise or NPS area-under the curve (r=0.37, p=0.26), high CTDI (r=0.41, p=0.21) and edge visibility of 1% target (r=0.59, p=0.05).This confirms that optimising contrast, resolution and noise are important for optimal image quality thus facilitating accurate and time saving contouring. Reconstruction kernel, FOV and noise level from X-ray tube current and tube rotation time were identified as possible parameters for adjustment to achieve this optimisation. The ranking of the results from the Catphan 604 phantoms were similar to those obtained from bespoke phantoms indicating suitability for future use. The assessment of clinical image noise identified that most scanners delivered consistent noise levels for different sized patients. This was not the case where fixed tube current was used for H&N patients giving high noise for larger patients or for one scanner (for prostate scans) which gave lower noise for large patients, requiring further investigation.

Conclusion:

A methodology using phantom metrics has been developed which can be used to assess image quality in a multicentre intercomparison. This methodology correlates well with clinical assessment of the suitability of images for radiotherapy contouring. Results can identify the poorest performing CT scanners and provide recommendations for parameter adjustment to improve image quality. The range of results obtained confirms scan protocol optimisation is necessary in some centres.

Keywords: CT, image, optimisation

References: