Purpose: Fluoroscopic image quality intrinsically contains a temporal component from lag and frame averaging. Lag can artificially increase the spatial DQE by decreasing noise in subsequent frames. Temporal performance of clinical fluoroscopy systems can be characterized through temporal modulation transfer function (TMTF) and temporal noise power spectrum (TNPS) using a rotating edge phantom. The goal of this study was to assess the reproducibility of the method using image sets acquired under identical conditions.
Methods: The experiment was conducted using a Philips Alura FD 20 interventional system. Images of rotating fan blades were acquired using normal and low dose fluoroscopy, as well as cine images. Static fan images facilitated spatial MTF measurements, which were necessary for isolating the TMTF from the spatio-temporal MTF. Three acquisitions of each type were taken to estimate measurement variability. TNPS measurements were made using open field images. To quantify the similarity of each trial, the root mean square error (RMSE) of each measurement was calculated against the average of all trials for each exposure type, then normalized to the mean value of the measured TMTF.
Results: The average TMTF normalized RMSE for normal and low dose fluoro, and cine were determined to be 0.009, 0.007, and 0.003, respectively. The low RMSE confirms the repeatability of this method. The TNPS results also exhibited low error, with normalized RMSEs ranging within 0.06-0.13.
Conclusion: The results indicate a high degree of precision in the method of calculating of the TMTF and TNPS, and provide reassurance that this method could be implemented into annual testing as a means of monitoring fluoroscopic system temporal performance. It would be possible to use these metrics to calculate the detectability of a given object function and track resultant detectability index values regularly.