Quantitative assessment of signal and noise properties of mammography digital detectors

Abstract

Recent measurements introduced in quality control protocols indicate large variations in the evaluated parameters, pointing out as their main cause the different detectors used. The objective of this work was to characterize, in terms of signal transfer property (STP) and noise component analysis (NCA), six mammography digital detectors at different DAK levels for a variety of beam qualities. A Carestream EHR-M3 CR detector and Siemens Opdima, Planmed Clarity, GE Essential, GE Pristina and GE Crystal Nova DR detectors were characterized in terms of STP and NCA, as a function of detector air kerma (DAK) at the X-ray detector input plane. An extended DAK range was used. A PTW Unidos E electrometer with PTW Freiburg ionization chamber were used to measure DAK. Attenuated beam qualities (obtained by adding 2 mm Al to the corresponding beam qualities) used were 28 kV with anode/filter combination Mo/Mo for the EHR-M3, Opdima, Essential and Pristina detectors; 28 kV with W/Rh for the Crystal Nova; and 28 kV with W/Ag for the Clarity. To study the influence of beam quality, 28 kV with Mo/Rh and 34 kV with Rh/Ag were also used for EHR-M3 and Pristina, respectively. STP measures the relationship between the input (DAK) and the output (pixel value) detector signals. NCA gives the fraction of total noise of each of its components and was calculated from the variance in mean pixel value. Noise stationarity was evaluated using maps of standard deviation across the image plane. Parameters were determined under stipulated (geometric and irradiation) conditions given in EUREF quality control protocol. Computed radiography (CR) detector showed a logarithmic response and digital radiography (DR) detectors showed a linear response in the tested DAK range (R2 > 0.999). For the DR detectors, the response function was also influenced by the beam quality, with a higher slope for more energetic beams. NCA showed that Opdima and EHR-M3 detectors are quantum limited below 290 μGy and 650 μGy, respectively. For higher DAK values, structure noise is the dominant noise source. All other detectors are quantum limited in the evaluated DAK range. Noise as a function of position in the image showed limited noise stationarity. The findings show that the signal transfer and noise characteristics are influenced by the properties of the detection system. Moreover, STP was influenced by the beam quality. NCA showed the influence of electronic and structure noise at lower and higher DAK, respectively. These results are useful for detector characterization.