A calibration approach to glandular tissue composition estimation in digital mammography

Abstract

The healthy breast is almost entirely composed of a mixture of fatty, epithelial, and stromal tissues which can be grouped into two distinctly attenuating tissue types: fatty and glandular. Further, the amount of glandular tissue is linked to breast cancer risk, so an objective quantitative analysis of glandular tissue can aid in risk estimation. Highnam and Brady have measured glandular tissue composition objectively. However, they argue that their work should only be used for "relative" tissue measurements unless a careful calibration has been performed. In this work, we perform such a "careful calibration" on a digital mammography system and use it to estimate breast tissue composition of patient breasts. We imaged 0%, 50%, and 100% glandular-equivalent phantoms of varying thicknesses for a number of clinically relevant x-ray techniques on a digital mammography system. From these images, we extracted mean signal and noise levels and computed calibration curves that can be used for quantitative tissue composition estimation. In this way, we calculate the percent glandular composition of a patient breast on a pixelwise basis. This tissue composition estimation method was applied to 23 digital mammograms. We estimated the quantitative impact of different error sources on the estimates of tissue composition. These error sources include compressed breast height estimation error, residual scattered radiation, quantum noise, and beam hardening. Errors in the compressed breast height estimate contribute the most error in tissue composition--on the order of +/-7% for a 4 cm compressed breast height: The spatially varying scattered radiation will contribute quantitatively less error overall, but may be significant in regions near the skinline. It is calculated that for a 4 cm compressed breast height, a residual scatter signal error is mitigated by approximately sixfold in the composition estimate. The error in composition due to the quantum noise, which is the limiting noise source in the system, is shown to be less than 1% glandular for most breasts.

Improvement in risk prediction, early detection and prevention of breast cancer in the NHS Breast Screening Programme and family history clinics: a dual cohort study

“…[35][36][37] Therefore, to improve the accuracy of the best performing risk model in each population of women (family history and general population), it is necessary to incorporate MD. We used a number of different density assessments to determine which adds most to the precision of breast cancer risk estimation.…”

Wavelet-based noise-model driven denoising algorithm for differential phase contrast mammography

“…To image different objects, in principle, f s 1 should be measured and calibrated with different materials and sample thicknesses in advance. Nonetheless, our major interest here is mammography, and for imaging breast samples, the beam hardening effect is not pronounced, as presented in [20] and as we can grasp from the similarity between f r 1 and f s 1 . Breast samples are usually assumed to have a simple composition (fat and glandular tissue, which have close attenuation properties), therefore f s 1 is estimated with only one material in our case.…”

Volumetric breast density measurement: sensitivity analysis of a relative physics approach

“…• Using a calibration device on a daily basis to calibrate the physics, 23 but such techniques still rely heavily on accuracy of the physics parameters such as breast thickness on an image-by-image basis.…”