Methodology development for breast cancer diagnosis using dual energy X-rays and digital tomosynthesis
Loading...
Date
Authors
Κούκου, Βάια
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Breast cancer screening and diagnosis in X-ray mammography rely on the detection and visualization of microcalcifications (μCs) and masses. The detection and visualization of μCs are relatively easy over a uniform tissue background. However, visualization of μCs could be obscured in mammograms by overlapping tissue structures. Dual energy (DE) imaging techniques can suppress the contrast between adipose and glandular tissues improving the detectability of μCs and masses. The current thesis is focused on the study of DE X-ray method applied on breast imaging, in order to improve the detection of breast calcifications.
A simulation study was initially performed, based on analytical modeling, incorporating a modified radiographic X-ray unit combined with a high resolution CMOS sensor (pixel pitch of 22.5 μm). The calcification contrast-to-noise ratio of the DE subtracted images (CNRtc) was calculated for various skin entrance doses and calcification thicknesses using monoenergetic beams and polyenergetic X-ray beams. An experimental DE method was followed based on the proposed simulation exposure conditions. Custom-made homogenous and inhomogeneous breast phantoms and two different μC phantoms, as well as, the ACR mammography accreditation phantom were used. Post-process noise reduction was applied on the DE images in order to increase the contrast-to-noise ratio of the DE subtracted images (CNRDE). Further evaluation of the DE method was included using DE images obtained from a formalin-fixed breast cancer specimen. Also, three different μC types were compared using the proposed DE method. Finally, the potential to better visualize μCs using DE in digital breast tomosynthesis (DBT) was investigated. The TORMAM phantom was imaged and log-subtraction algorithm was applied to each pair of low-energy (LE) and high-energy (HE) slices after reconstruction.
The various examined experimental setups demonstrated increased visibility of the μCs using the proposed DE method. Moreover, the DE images of breast specimen were directly comparable to the mammographic image and increased μC information was identified. The improvement in visualization of μCs, in the current method, is attributed to the use of higher kV X-rays from a modified radiographic unit with heavy filtering leading to larger spectral separation, while preserving MGD in acceptable levels. The use of post-processed DE images resulted in CNRDE values approximately 3 times that of the original DE image. In DE DBT images the μCs clusters were better defined than in the corresponding LE images. The initial results showed that DE DBT imaging could be a promising modality.
Description
Keywords
Dual energy, Breast imaging, X-rays, Calcifications, Digital breast tomosynthesis, Post-processing, Contrast-to-noise ratio (CNR)