Adaptive image interpolation for full-field digital x-ray mammography

Area of Science:

• Medical Imaging
• Digital Mammography
• Image Processing

Background:

• Full-field digital x-ray mammography systems often utilize multiple adjacent 2D detectors.
• Gaps between these detectors result in image seams, compromising the integrity of the composite mammogram.
• Accurate reconstruction of data in these seams is crucial for effective diagnostic interpretation.

Purpose of the Study:

• To develop and evaluate an adaptive linear interpolation method for seamless image reconstruction in digital mammography.
• To address the challenge of missing data caused by detector gaps in multi-detector mammography systems.
• To improve the quality of composite mammograms by minimizing artifacts from detector seams.

Main Methods:

• An adaptive linear interpolation technique was developed to estimate missing data within image seams.
• The interpolation path is dynamically determined by analyzing the similarity of boundary profiles between adjacent subimages.
• The method was tested using both phantom images and clinical mammograms.

Main Results:

• The proposed adaptive linear interpolation method demonstrated significantly superior performance compared to conventional linear interpolation.
• Experiments showed effective estimation of missing data in the seams, leading to reduced image artifacts.
• Improved image quality was observed in both phantom and clinical datasets.

Conclusions:

• Adaptive linear interpolation is a highly effective technique for reconstructing images from multi-detector digital mammography systems.
• This method significantly enhances image quality by accurately filling data gaps caused by detector seams.
• The findings suggest a valuable improvement for digital mammography image processing and diagnostic accuracy.