ented. into account when processing algorithms are

ented.  In  other words,  characteristic features of cancer have  to  be  displayed with optimal contrast to avoid being missed or  misinterpreted.  Within the limited possibilities of  conventional screen-fi lm  mammography,  this matching problem has received much attention in the past.  Many  innovations have  been made  in mammography  to  enhance  visibility  of  cancers. Most notable in this respect is the gradual increase of  contrast in the interior  region  of  the breast at  the cost of  contrast in the periphery,  where cancer seldom occurs.  This change was only  possible due to  development  of  more  accurate  automatic exposure control (AEC) devices. With the latest generation  of  mammography fi lms, the skinline and large parts of  the periphery are hardly visible on  mammography  fi lm  alternators. With the introduction  of  digital mammography, a wide range of  new possibilities has become available to enhance  mammograms. An overview of  digital mammogram processing techniques will  be  presented here. techniques was given by  Pisano  et al. (2000a). Here, we will  discuss some of  the basic methods currently employed and more  advanced methods that  will  likely become available in the next generations.  Basic processing methods include grayscale transforms  and adaptive contrast enhancement.  A common  dedicated mammogram processing method is peripheral enhancement,  which has been adopted widely  by  manufacturers to  overcome shortcomings of  the dynamic range of digital displays. 5.2 Grayscale Transforms Digital image processing is only  one part  of  the mammographic imaging chain. Ideally, the design  of a medical image processing method should be  independent of  image acquisition and display.  For image display,  this ideal can be  achieved by  ensuring that display devices conform to  the DICOM display standard.  Therefore,  it is important that  one is aware of the mechanisms  used in the defi nition  of  this standard.  Mappings  to  convert pixel values to  luminance in display devices depend on  parameter settings  in the processed images,  like  window/level settings  and values of  interest lookup  table (VOI LUT).  Selection of  appropriate values of  these parameters is an issue that  should be  addressed by  processing algorithms. On the other end of  the chain is the acquisition device.  Digital detectors in mammography  devices differ and these differences have  to  be  taken into account when processing algorithms  are designed. Examples are variation in image resolution,  gain, modulation  transfer function, and noise characteristics.  Fortunately,  important acquisition parameters such as anode material,  ltration,  and kVp are provided in the DICOM header and can thus be  used in processing algorithms.  Despite many differences, there is a major advantage in the use of  digital detectors:  in the range of  interest,  pixel values are more  or less proportional to  X-ray exposure at  the detector. This allows design  of  robust processing algorithms which can be  applied to  a variety of  systems. Digital mammography  manufacturers  have  only just begun to  explore  the enormous benefi t that  digital processing and display may provide. An interesting pictorial essay of  some mammographic processing