With strong detection and classification abilities, mammography and other forms of breast imaging are among the most important tools in fighting breast cancer. But with these strengths come several challenges-namely, ensuring minimal degradation in imaging quality caused by device limitations.
Because improving screening and diagnostic accuracy in digital mammography is difficult, breast imaging display systems push the limit of display technology in several aspects of performance. On the positive side, mammography display systems can have exquisite resolution (spatial and grayscale), fairly wide viewing angles and low noise. But the scientific question remains: How can we best display large pixel array images with subtle contrast and small details for correct diagnoses?
Liquid crystal display (LCD) technology can deliver several million pixels for displaying a digital mammogram with a 1-to-1 presentation mode between detector and display pixel. Pixels are controlled by thin-film transistors that allow more or less light to be transmitted through a panel, modulating backlight intensity. This circuitry can be used (through improved gray level-to-luminance tables) to correct for nonuniformities in the screen, and reduce unwanted pixel variations.
As to how many shades of gray are needed, evidence suggests that greater than 8-bit palettes are beneficial for breast X-ray image interpretation. Two techniques in particular help overcome the standard driver limitation of 256 gray shades to achieve 10- and 11-bit grayscales: Subpixel averaging addresses fractional areas of a pixel (corresponding to color subpixel areas), although it compromises spatial resolution. And temporal averaging rapidly modulates pixel luminance over many frames, although it hampers temporal response.
In the reading room, ambient light greatly affects the quality of a displayed image. The magnitude of degradation is associated with a display's reflection properties as well as the directionality of light sources and reflectors in the room. Diffuse ambient light needs to be accounted for when calibrating a device's luminance, ensuring that contrast of dark screen areas is higher than ambient light luminance to prevent loss of detail in dark areas.
Image quality is also susceptible to hardware performance over time. Periodic monitoring of display quality and calibration is essential, and several professional groups, including the American Association of Physicists in Medicine (AAPM) and American College of Radiology (ACR), have developed guidance documents for acceptance testing and monitoring medical imaging displays. The frequency and rigor of testing is still under debate, but evidence indicates that sporadic and unsystematic quality control leads to significant decreases in performance, affecting diagnostic accuracy and increasing reader variability. New techniques to automate calibration and testing should prove useful for all-size clinics where display quality requires monitoring and maintenance.
In the future, adding color filters to a monochrome LCD design may yield color displays for breast imaging applications without significant degradation in most performance areas. However, we must evaluate noise introduced by display panel variations and the presence of subpixels. (In some products, noise-reduction techniques have been successful.)
Display component challenges will be numerous with 3-D breast X-ray imaging applications, such as tomosynthesis, stereomammography, CT and multimodality imaging. Systems will need reliable grayscale performance, possible color overlays as with other organs such as the chest and abdomen, and additional features for data visualization. Moreover, 2-D projection images will progressively incorporate 3-D reconstructions in a cine mode or volume rendering of segmented data for browsing through slices. Which display protocol delivers the highest interpretation efficiency and accuracy is yet to be determined.
New requirements will be needed for color and temporal characteristics. It is known, for example, that poor temporal LCD response is the result of slow liquid crystal molecule response and hold-type (frame-at-a-time) addressing schemes. This affects exact mapping, and degrades image quality in flythroughs of large volumetric data sets and contrast rendition in fast sequence stack-mode browsing.
Furthermore, improvements in angular, temporal and noise performance are needed. Visualization of large, multidimensional image datasets generated by breast CT or tomosynthesis might overwhelm existing reading paradigms and require new display methods, perhaps representing 3-D information in stereoscopic or true 3-D form.
Aldo Badano, PhD, is lab leader for the imaging physics laboratory, CDRH/NIBIB Laboratory for the Assessment of Medical Imaging Systems, division of imaging and applied mathematics, Office of Science and Engineering Laboratories, FDA Center for Devices and Radiological Health in Silver Spring, Md. The mention of commercial products herein is not to be construed as an actual or implied endorsement of such products by the Department of Health and Human Services. This is a contribution of the FDA and is not subject to copyright.