It is widely recognized that high-quality images are imperative for the reliable detection and accurate characterization of subtle lesions in the breast with mammography. The quality of the images depends critically on the design and performance of the radiographic unit and image receptor and on how that equipment is used to acquire and process the mammogram. In addition, the type of display and the conditions under which the image is viewed have an important effect on the ability of the radiologist to extract the information recorded in the mammogram.<sup>31Kundel H.L., Revesz G. Lesion conspicuity, structured noise, and film reader error AJR Am J Roentgenol 1976 ;  126 : 1233

Haga clic aquí para ir a la sección de Referencias</sup> In this article, some factors affecting image quality and radiation dose to the breast in screen-film and digital mammography are discussed.

In general, the phrase mammographic image quality can be considered to indicate the clarity with which radiologically significant details can be perceived in an image. In turn, high mammographic image quality should contribute to high performance in detecting and diagnosing breast cancer. There is, however, no well-defined standard for specifying mammographic image quality. The relationship between physical properties of the radiographic image (such as contrast, resolution, and noise) and the ability of the observer to properly detect and interpret relevant image features is not well understood.<sup>47Vyborny C.J., Schmidt R.A. Technical image quality and the visibility of mammographic detail Syllabus of Categorical Course on Technical Aspects of Mammography Oak Brook, IL: Radiological Society of North America (1994).  103

Haga clic aquí para ir a la sección de Referencias</sup> Currently, probably the most effective tool for inferring this is receiver operating curve methodology,<sup>33Metz C.E. Applications of ROC analysis in diagnostic image evaluation The Physics of Medical Imaging: Recording System Measurements and Techniques New York: American Institute of Physics (1979).  546

Haga clic aquí para ir a la sección de Referencias</sup> applied retrospectively to clinical images where the true disease state is known. Receiver operating curve testing can be used to assess the overall performance of a radiologist in combination with a particular imaging system at detecting or diagnosing breast cancer in terms of sensitivity at varying levels of specificity. This creates a measure that is based on perception of information in the mammogram and is essentially independent of the level of conservatism of the radiologist in calling abnormal findings. Unfortunately, receiver operating curve studies are complex, time-consuming experiments, requiring large image data bases with known truth data regarding disease. They require many image readings by many observers, and are often not practical for routine measurement of image quality.

Logically, mammographic image quality should be related to certain image attributes that can be described technically, such as spatial resolution, contrast, image noise and signal-to-noise ratio (SNR), and the absence of artifacts.<sup>47Vyborny C.J., Schmidt R.A. Technical image quality and the visibility of mammographic detail Syllabus of Categorical Course on Technical Aspects of Mammography Oak Brook, IL: Radiological Society of North America (1994).  103

Haga clic aquí para ir a la sección de Referencias</sup> It is accepted that these are important parameters that affect the ability to detect or characterize microcalcifications or to visualize fine fibrillar structures radiating from a mass or the presence of architectural distortion. The exact relationship, however, is not yet well understood (i.e., we still do not know how to define what constitutes optimal or necessary quality for diagnostic accuracy in terms of technical parameters). At present, about the best we can do is to attempt to correlate differences in receiver operating curve performance with differences in the technical aspects of image acquisition and display. This issue is the subject of continuing research.<sup>4Bencomo J.A., Haus A.G., Paulus D.D., y al. Method to study the effect of controlled changes of breast image total system modulation transfer function (MTF) on diagnostic accuracy Proc SPIE 1982 ;  347 : 84

Haga clic aquí para ir a la sección de Referencias, 5Bunch P. Advances in high-speed mammographic image quality Proc SPIE 1999 ;  3659 : 120

Haga clic aquí para ir a la sección de Referencias, 7Bunch P.C., Hamilton J.F., Sanderson G., y al. A free response approach to the measurement and characteristics of radiographic observer performance Proc SPIE 1977 ;  127 : 124

Haga clic aquí para ir a la sección de Referencias, 10Chan H.P., Vyborny C.J., McMahon H., y al. Digital mammography ROC studies on the effects of pixel size and unsharp-mask filtering on the detection of subtle microcalcifications Invest Radiol 1987 ;  22 : 581 [cross-ref]

Haga clic aquí para ir a la sección de Referencias</sup> Nevertheless, there is a strong correlation between radiologists' rejection of mammograms as having inadequate quality and low measured values of resolution, contrast, and SNR or the excessive prevalence of artifacts.

In many cases the optimum values of these parameters are not all simultaneously achievable, at least not at a reasonable radiation dose, and tradeoffs must generally be considered. The optimum compromise between technical parameters in forming the image to achieve high mammographic quality is likely to be task dependent.<sup>46Vyborny C., Schmidt R.A. Mammography as a radiographic examination: An overview Radiographics 1990 ;  9 : 723

Haga clic aquí para ir a la sección de Referencias</sup> For example, the image characteristics required to allow detection of a large lesion in a fatty breast can be very different from those needed to visualize microcalcifications in a rather dense breast.

The following discussion considers those factors that can be quantified objectively. We use the term technical image quality to include those factors that are measurable in the imaging process rather than the variables entering into interpretation of the image.