Quantifying the impact of respiratory-gated 4D CT acquisition on thoracic image quality: a digital phantom study

Prospective respiratory-gated 4D CT has been shown to reduce tumor image artifacts by up to 50% compared to conventional 4D CT. However, to date no studies have quantified the impact of gated 4D CT on normal lung tissue imaging, which is important in performing dose calculations based on accurate estimates of lung volume and structure. To determine the impact of gated 4D CT on thoracic image quality, the authors developed a novel simulation framework incorporating a realistic deformable digital phantom driven by patient tumor motion patterns. Based on this framework, the authors test the hypothesis that respiratory-gated 4D CT can significantly reduce lung imaging artifacts.

Our simulation framework synchronizes the 4D extended cardiac torso (XCAT) phantom with tumor motion data in a quasi real-time fashion, allowing simulation of three 4D CT acquisition modes featuring different levels of respiratory feedback: (i) "conventional" 4D CT that uses a constant imaging and couch-shift frequency, (ii) "beam paused" 4D CT that interrupts imaging to avoid oversampling at a given couch position and respiratory phase, and (iii) "respiratory-gated" 4D CT that triggers acquisition only when the respiratory motion fulfills phase-specific displacement gating windows based on prescan breathing data. Our framework generates a set of ground truth comparators, representing the average XCAT anatomy during beam-on for each of ten respiratory phase bins. Based on this framework, the authors simulated conventional, beam-paused, and respiratory-gated 4D CT images using tumor motion patterns from seven lung cancer patients across 13 treatment fractions, with a simulated 5.5 cm(3) spherical lesion. Normal lung tissue image quality was quantified by comparing simulated and ground truth images in terms of overall mean square error (MSE) intensity difference, threshold-based lung volume error, and fractional false positive/false negative rates.

Averaged across all simulations and phase bins, respiratory-gating reduced overall thoracic MSE by 46% compared to conventional 4D CT (p ∼ 10(-19)). Gating leads to small but significant (p < 0.02) reductions in lung volume errors (1.8%-1.4%), false positives (4.0%-2.6%), and false negatives (2.7%-1.3%). These percentage reductions correspond to gating reducing image artifacts by 24-90 cm(3) of lung tissue. Similar to earlier studies, gating reduced patient image dose by up to 22%, but with scan time increased by up to 135%. Beam paused 4D CT did not significantly impact normal lung tissue image quality, but did yield similar dose reductions as for respiratory-gating, without the added cost in scanning time.

For a typical 6 L lung, respiratory-gated 4D CT can reduce image artifacts affecting up to 90 cm(3) of normal lung tissue compared to conventional acquisition. This image improvement could have important implications for dose calculations based on 4D CT. Where image quality is less critical, beam paused 4D CT is a simple strategy to reduce imaging dose without sacrificing acquisition time.