Correlation between exposure (mAs) and image quality for a rapid cone-beam CT on a ring gantry linear accelerator

Abstract

Background: The adoption of HyperSight CBCT has become widespread due to its rapid CBCT scan acquisition (< 6 s) and superior image quality. Numerous studies have been conducted to evaluate HyperSight CBCT's image quality and the accuracy of treatment planning dose calculations using HyperSight CBCT. Nevertheless, for the main function of CBCT as a tool of image guided radiation therapy (IGRT), one clinically significant question remains unanswered: What is the quantitative correlation between exposure level (mAs) and desired image quality? This study aims to explore the quantitative correlation between exposure level and image quality, providing guidelines for selecting the appropriate CBCT mAs based on imaging objectives and patient size for the optimal IGRT alignment.

Purpose: To investigate the correlation between HyperSight CBCT (equipped on Halcyon linear accelerator) exposure level and image quality.

Methods: Nine-series of 261 sets CBCT scans were acquired using HyperSight on the Halcyon (125 kV, 133-971 mAs, FOV: 28.2, 36, and 53.8 cm). Three-series were acquired with the CatPhan 604 alone, three-series with CatPhan 604 surrounded by a 32 cm custom-made acrylic annulus, and three series with CatPhan 604 surrounded by a 40 cm custom-made acrylic annulus. Image quality metrics were analyzed included signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), image noise, low-contrast visibility (LCV), uniformity index (UI), integral non-uniformity (INU), and modulation transfer function (MTF). The correlation between mAs and image quality was evaluated.

Results: The SNR and CNR for all material inserts increased with rising mAs for all scans, with a higher amplitude of increase for scans without annuli. The SNR and CNR were higher for scans without annuli. Both image noise and LCV decreased with rising mAs and were higher for scans with annuli. The UI was independent of mAs for scans of CatPhan alone and CatPhan with a small annulus, and was slightly higher for scans without annuli (∼0.1 vs ∼0.2). The INU slightly increased with rising mAs and was higher for scans without annuli. MTF was independent of mAs and with or without annuli. MTF values varied with FOV, with smaller FOV yielding better image sharpness. All parameters were FOV-dependent except INU, and generally, the larger the FOV, the better the image quality (except spatial resolution).

Conclusions: HyperSight CBCT acquisition is recommended to be performed with a maximum FOV of 53.8 cm. To meet the Rose criterion, for a 20 cm width object, CBCT should be acquired at ≥154 mAs; for a 32 cm width object, CBCT should be acquired at ≥572.6 mAs to reach a 3σ confidence level; for a 40 cm object, CBCT should be acquired at ≥572.6 mAs to reach a 3σ confidence level for all materials except polystyrene (soft-tissue).

Keywords: HyperSight CBCT; Rose criterion; image quality.