Gisella Gennaro PhD, Presenter: Nothing to Disclose

Gilberto Contento PhD, Abstract Co-Author: Nothing to Disclose

Fulvio Fornasin PhD, Abstract Co-Author: Nothing to Disclose

Cosimo di Maggio MD, Abstract Co-Author: Nothing to Disclose

Automation of long-term reproducibility tests in digital mammography

A software package was developed to performed automatic quality controls from phantom images obtained by digital mammography systems. The software tool was validated for the TORMAS mammographic phantom. Image datasets were acquired by daily exposing the phantom with a GE Senographe 2000D digital unit and a Fuji CR system, to evaluate the long-term reproducibility of both DR and CR systems. Raw images were available for the DR system, while both unprocessed and post-processed images were used for the CR system. Many quantitative image quality (IQ) indices, contrast-to-noise ratio (CNR) of circular details, stepwedge contrast, mean pixel value (MPV) and standard deviation of the background, were automatically extracted, together with detail counts determined by using specific threshold values.

Quantitative analysis showed sufficient sensitivity to relate fluctuations in exposure parameters to variations in image quality indices. In comparison, detail counts were found less sensitive in detecting image quality changes. The quality controls over two months showed a variation in IQ quantitative indices consistent with the variation of exposure parameters, and lower than 3% for the DR system. For the CR system, the results showed that the absence of raw data prevent the detection of input fluctuations from image analysis; in fact, despite the mAs variation close to 3%, the MPV of the larger details keeps within 1%. By removing any processing from the CR scanner, the scanner variability could be derived, lower than 2%. Detail counts showed greater variation for both DR and CR systems, between 6 and 18%.

Automatic quality controls with quantitative analysis of multiple IQ indices, have demonstrated to be effective in monitoring stability in digital mammography, when raw images are available. Detail count seems not be a reliable and effective method, despite the automation of image analysis.

Automated quality controls can be successfully performed in digital mammography. The sensitivity in evaluating equipment stability is superior for DR systems, for which raw images can be obtained.

Gennaro, G, Contento, G, Fornasin, F, di Maggio, C, Automatic Quality Controls in Digital Mammography.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5016080.html