A practical method to calibrate and optimise automatic exposure control devices for computed radiography (CR) and digital radiography (DR) imaging systems using the signal-to-noise ratio (SNR) metric

Abstract

We have established in recent virtual clinical trials that signal-to-noise ratio (SNR) is the most appropriate metric for calibrating AEC devices used with digital imaging systems for chest, abdomen, pelvis and spine radiography. However, the practical calibration of such devices is not straightforward. The purpose of this study was to expand on previous work by developing a practical method for AEC calibration using the validated SNR metric that may be used by medical physicists in the field for phosphor plate radiography (CR) and direct digital radiography (DR - wireless and integrated) imaging systems. This methodology used a uniform tissue equivalent phantom (Poly-methyl methacrylate (PMMA)) so comparison was made with an installation vendor’s method of using copper plates. System transfer properties (STP) of each system were derived to linearise all images required for analyses. All STP correction was done in real time. The relationship between detector air kerma (DAK) and STP corrected SNR was then derived at four tube voltages (60, 80, 100 and 120 kVp) and using this relationship, together with the target value of the calibration metric (SNRtarget), it was possible to calculate the DAK required at each tube voltage to derive a calibration curve for each system; CR systems exhibited a linear increase with kVp, DR systems exhibited a ‘U’ shape. The curve derived using the vendor’s method was not considered optimised because it did not hold SNR constant. This work has demonstrated that a relatively simple method to calibrate AEC devices, using an easily accessible tissue equivalent phantom, can be used in the field in real time by medical physicists working with installation engineers. AEC calibration curves that produce clinically adequate image quality with acceptable patient dose have been installed in our radiology department for CR and DR systems using the methodology described in this paper.