Room: Exhibit Hall | Forum 9
Purpose: To establish a method to select treatment-appropriate breathing phases based on analysis of respiratory waveforms acquired during 4DCT scans.
Methods: 4DCT waveforms acquired using the Varian RGSC system during CT simulation of 15 patients with lung or abdominal lesions were retrospectively analyzed. Specifically, for each patient, the anterior-posterior amplitude signal of an external surrogate marker block as a function of time was extracted. Phase-binning of the amplitude signal was then performed and the statistical variance in amplitude as a function of phase was calculated. By taking the derivative of the amplitude vs. phase data, the velocity as a function of phase was computed; a moving average mean was used to de-noise the resulting data. Positive and negative velocities represented motion toward maximum inspiration and toward maximum expiration, respectively. Velocity thresholds of Â±50% of the maximum positive and negative velocities were established. Subsequently, the largest contiguous phase range for which the mean velocity signal fell within the thresholds was determined. This range was compared with the range selected for treatment based on physicians' analysis of target motion in 4DCT scans.
Results: The phase ranges selected using the velocity-based approach were, on average, comparable to those selected using 4DCT analysis. The selected phases also correlated well with local minima in the variance of the amplitude signal as a function of phase.
Conclusion: For the patients studied, since velocity-based selection of gating phases correlated well with phases chosen using 4DCT-based analysis of target motion (the standard-of-care), the motion of the external surrogate seems to accurately reflect internal target motion; it may therefore be used as a verification of 4DCT-based phase selection. Furthermore, in the event that a patient simulated with phase gating needs to be treated with amplitude gating, this technique could be used to establish equivalent amplitude thresholds without requiring re-simulation.