Film screen mammography is effective in the detection of early breast cancer and in reducing breast cancer mortality. Although film screen mammography has been improved considerably, several technical factors limit the ability of mammography to display fine abnormalities at an efficient radiation dose [1]. Film mammography serves as the image acquisition device as well as the storage and display devices. With digital mammography these three functions are separated, and each function can therefore be more easily optimized.
Recently, a digital unit in the United States has been approved by the Food and Drug Administration for full breast imaging. Prior to this approval, spot fields for stereotactic biopsy guidance and for the diagnostic evaluation of breast lesions (i.e., spot magnification views) were performed with direct digital techniques.
In direct digital mammography, the image is acquired on a digital detector and is displayed on a workstation or laser printed onto film as a two—dimensional matrix of pixels. The image data may be stored or archived on optical disks as well as stored as laser printed images. Key performance specifications of a digital mammography system relate to spatial resolution and contrast sensitivity [2].
The resolution of a digital image is dependent on the size of the pixels and their spacing. For film/screen mammography the spatial resolution is approximately 20 lines pairs/mm. A digital system would need to have a 25 µm pixel size to replicate this level of resolution. For the display at this resolution of the images from a 24 cm x 30 cm size receptor, as is typically used in mammography, the matrix would be 9600 x 12,000 pixels. At this time, there is no workstation available that can display the whole breast image of this size at full resolution.
The actual resolution requirements for digital mammography remain unknown. A 25 µm pixel size (~ 20/lp/mm) is probably not necessary, and the actual pixel size and spatial resolution may be in the range of 50—100 µm [2]. This would obviously facilitate both the development of detectors at a more reasonable cost and the display of the images on workstations. Studies have shown improved detectability of low contrast objects on digital systems at a resolution less than that of film systems (at 50—100 µm pixels which correlates to 10 or 5 line pairs/mm) [3, 4].
Digital mammography has a much higher dynamic range than film/screen mammography, allowing for better contrast over a wider range of intensities and a lower noise level. The potential advantages of digital systems are many, including the improved visualization of abnormalities in dense breasts and the possibility of post—processing the images, which can improve lesion characterization. Post—processing can be used to enhance contrast and magnify fine calcifications, as well as for image inversion and filtering. A digital image may be coupled to telemammography, thereby increasing access to expert interpretation for underserved areas. The coupling of computer—aided diagnosis and digital mammography provides the potential for double readings of screening mammography, and the possibility of using artificial neural networks to assist in lesion characterization.
Great challenges exist in the development of workstations with user—friendly display protocols that are also capable of displaying full breast images at adequate resolution. Such usable display protocols would allow the radiologist to rapidly move from both MLO views to the CC and MLO view of one breast to the prior and current views, for example. Another option is to laser print onto film the digital image, but this removes the ability to use interactive post processing tools on a workstation (i.e., window/level magnification or inversion) to optimize viewing of a region of interest.
Various types of direct digital systems are under consideration and development:
- Digital Camera
The coupling of a phosphor screen to a photodetector or charged—coupled device (CCD) that is substituted for film. The CCD converts light received from the phosphor into electrons. The challenge of this technology is that the photodetectors are on single microchips and currently the largest one is only 5x8 cm, much smaller than the full breast size. Therefore, for imaging of the whole breast, demagnification of the phosphor image through lenses or fiberoptic tapers is used to couple the information to smaller CCD's. Other systems utilize linked or "stitched" CCD's to cover the entire region of interest, but this can result in "stitching artifacts" that may mar the image.
- Scanned Beam Detectors (point, line or slot scanners)
A small—area photodetector in conjunction with the radiation beam is scanned over the breast, creating the full image. This technology may be associated with increased radiation exposure, as well as a blurred image from the prolonged exposure time.
- Photostimulable phosphors (Computed Tomography)
These plates are used to capture the image of the breast and are laser—scanned to collect information and the images are interpreted as laser prints. The resolution of these systems is limited to about 4—5 lp/mm, which may not be adequate for mammography.
- Amorphous selenium
An electrostatic charge representing the image rests on the surface of the plate and is read by scanning a laser beam over its surface. The inherent resolution of this type of system is better than that of the photostimulable phosphors.
- Amorphous silicon
A scintillator absorbs the x—rays and converts them to light. Highly sensitive diodes containing one pixel of information are deposited on a plate of amorphous silicon. The charge at each pixel is read out digitally and sent to an image processor.
As we move forward toward the digital age in mammography, we must consider and overcome several challenges. The display of images in a user—friendly format and at an acceptable resolution on high resolution workstations with appropriate display protocols is necessary. This will allow for post—processing and manipulation of the images. The management of patient throughput with potential increased time for interpretation and image retrieval must be balanced with the advantages of the digital mammography (image acquisition and archiving). Lastly, the cost of the equipment is significantly more than that of film screen systems, and the negative effect of cost on the implementation of telemammography in rural or small practices must be considered.
Mammography, with its inherent need to image fine detail at a resolution far greater than other radiologic examinations require, has remained the last bastion of analog imaging. Now with the development of the direct digital systems we may see the advantages of digital imaging in this subspecialty.
