Optimization of K-edge subtraction imaging using a pixellated spectroscopic detector

Conventional K-edge subtraction imaging is based around the acquisition of two separate images at energies respectively below and above the K-edge of a contrast agent. This implies increased patient dose with respect to a conventional procedure and potentially incorrect image registration due to patient motion. We present results obtained with a pixellated spectroscopic CdTe detector. A spectroscopic detector allows simultaneous acquisition of the two images by integrating appropriate bands from the transmitted X-ray spectrum, thus removing the above limitations; the photon counting capability of the detector limits the noise to statistical noise, thus minimizing the dose for a given signal-to-noise ratio. Furthermore, an appropriate choice of the integration bands allows optimization of image quality, resulting from a trade-off between background removal (maximum with a narrow band) and low statistical noise (achieved with a broad band). Results obtained with a custom test-object, simulating breast structures, and with an iodine-based contrast agent, are presented for two different image subtraction algorithms: logarithmic subtraction and dual-energy linear combination. Whilst being conceptually simpler, logarithmic subtraction is strongly dependent upon the position and width of the band selected, while linear combination allows better background removal even with a broad energy band, and therefore better image quality.