Dual-basis-material decomposition for dual-kVp cone-beam CT breast imaging

摘要

The image contrast among tumor tissue and non-tumor tissue in a breast volume reconstructed by cone-beam computed tomography (CBCT) breast imaging is usually very small. Striving for contrast enhancement by exploiting the energy-dependent x-ray attenuation, we herein report a dual-kVp CBCT breast imaging modality. Based on dual-kVp cone-beam scan and dual-basis-material decomposition, we represent breast tissues in a two-feature space that is spanned by two basis material functions. Though linear independent, the basis material functions are partially correlated or non-orthogonal. Therefore, the dual-material-equivalent decomposition is essentially a sort of non-orthogonal expansion, which is more useful for material classification than for quantitative measurement. On the other hand, the curse of dimensionality discourages high-dimensional feature space that may be spanned by using more basis material functions. For optimizing the material space, we suggest the use of two material spaces (different in basis functions). The first material space is spanned by {bone, polyethene}, which covers a wide range of x-ray attenuation, from calcification to soft tissue. The second material space is spanned by {teflon, fat}, which covers a small range, suitable for low-contrast soft tissue discrimination. Based on x-ray energy-dependent attenuation of materials and x-ray spectra, we simulated the dual-kVp CBCT breast imaging modality with pre-reconstruction scheme, thereby showing the feasibility for image contrast enhancement by dual-kVp technique. With the test materials: breast tissue, water, and soft tissue, the simulation showed that the dual-kVp technique could double the image contrast in this particular case. Through experiment with our CBCT prototype, we demonstrated dual-kVp volume subtraction, showing the energy-dependent x-ray attenuation.