Session Item

Respiratory signal-guided 4DCT scanning reduces image artifacts compared to conventional 4CT approaches, especially for irregular breathing. Intelligent 4DCT sequence scanning (i4DCT) is a corresponding step-and-shoot scan protocol that selects beam-on/off periods online such that the data sufficiency condition (DSC) is fulfilled at each couch position. In case of long expiration phases, however, each beam on period covers an end-exhalation state that may last longer than necessary. We propose and evaluate a novel dose modulation scheme to reduce the associated unnecessary dose to the patient.

Dose down- and up-regulation were defined rule-based, based on the patient's breathing signal and the representative breathing cycle learned before and during scanning: For dose down-regulation in prolonged exhalation, time (wait until expiration is reached), amplitude (below a pre-defined threshold) and phase (arc interval in phase space) criteria were to be fulfilled; up-regulation after exhalation (usually easier to detect) was defined by an amplitude criterion. The corresponding potential of dose reduction was evaluated in-silico (ie, by simulation of the scanning process) using 4DCT breathing signals of 167 patients (signals acquired by the Varian RPM system). Dose reduction was determined as the fraction of the down-regulated dose delivery time to the overall beam-on time. Assuming standard 10-phase 4DCT, periods at risk (‘fails’) were defined as breathing cycles for which the down-regulation period covered the entire phase-specific amplitude range for a specific breathing phase (ie, no appropriate reconstruction of the phase image possible for the corresponding beam-on period).

The proposed dose modulation strategy resulted in an average dose reduction of 11+/-5% per patient (maximum: 28%). As intended, down-regulation mainly affected the end-exhalation phase (91%) and adjacent expiration (6%) and inhalation (1%) phases. For the proposed rules and inherent algorithm parameters, ‘Fails’ were observed in 3% of the analyzed beam-on periods (66 of 1935 beam-on periods). The amount of dose reduction was not significantly correlated with the average breathing cycle length of the patient (Spearman’s rho 0.12, p=0.12), but the fraction of ‘fails’ was anti-correlated with the average cycle length (rho = 0.36; p<0.001). Among patients with the 10% longest average breathing cycles, the fraction of ‘fails’ was reduced to <1%.

The results support the potential for significant dose reduction in respiratory signal-guided 4DCT by online dose modulation. The reverse correlation of average breathing cycle length and the fraction of ‘fails’ could allow for the identification of patients that safely (ie, without loss of image quality) benefit from the proposed dose modulation scheme when extending the study to a larger cohort and adapting the applied dose modulation criteria parameters.