Image processing algorithms for digital mammography: a pictorial essay

Area of Science:

• Radiology
• Medical Imaging
• Image Processing

Background:

• Digital mammography systems utilize image processing algorithms to adjust image contrast.
• Various display algorithms exist, each with unique benefits and drawbacks for breast imaging tasks.

Purpose of the Study:

• To evaluate the advantages and disadvantages of different image processing algorithms in digital mammography.
• To assess the impact of these algorithms on lesion conspicuity, detail preservation, and screening performance.

Main Methods:

• Review of digital mammography display algorithms including manual intensity windowing, histogram-based windowing, mixture-model windowing, contrast-limited adaptive histogram equalization, unsharp masking, peripheral equalization, and Trex processing.
• Analysis of how each algorithm affects image contrast, lesion edge visibility, detail preservation, and potential for enhancing nuisance information.

Main Results:

• Manual windowing is operator-dependent. Histogram-based methods improve edge conspicuity but lose detail. Mixture-model enhances borders but may obscure dense areas.
• Contrast-limited adaptive histogram equalization can highlight edges but may increase false positives in screening. Unsharp masking sharpens borders but can alter lesion appearance.
• Peripheral equalization preserves peripheral details but can flatten non-peripheral contrast. Trex processing visualizes detail and edges but reduces overall contrast.

Conclusions:

• No single image processing algorithm is optimal for all digital mammography applications.
• Algorithm selection requires careful consideration of the specific diagnostic or screening task to balance lesion visibility with potential artifacts and information loss.