A REGIONAL CONVOLUTION KERNEL ALGORITHM FOR SCATTER CORRECTION IN DUAL-ENERGY IMAGES - COMPARISON TO SINGLE-KERNEL ALGORITHMS

Single kernel scatter correction algorithms are based on the model tha t the scatter field can be predicted by convolution of the primary int ensity (I-prim) with a spatially invariant scatter point-spread functi on (PSF). Practical limitations (I-prim unknown) suggest the substitut ion of the total detected intensity (I-det) for I-prim as the source i mage in the convolution. In regions of high scatter fraction (SF), I-d et is a poor approximation of I-prim, thereby causing an overestimatio n of scatter originating in the region. This contributes to errors in estimating detected scatter in the mediastinum and neighboring regions . A technique using a regionally variable point-spread function that s ignificantly reduces RMS error in estimation of the primary image as c ompared to the single PSF method is investigated. The regionally varia ble convolution method employs a larger PSF in the mediastinum and a s maller PSF in the lungs to reduce the error in estimating the scatter throughout the image. The method to allow for patient differences has also been expanded and various implementations of these methods have b een compared. Results show that the dual-kernel algorithm is always mo re effective than an equivalent single-kernel algorithm. The dual-kern el algorithm using a predicted scatter fraction curve gives an overall RMS error in the primary of as low as 20.8% which is equivalent to 8. 7% RMS error in the scatter. The dual-kernel method using a predicted scatter fraction curve approaches the accuracy of the single-kernel me thod using patient specific scatter measurements. Because using indivi dual scatter measurements is a less desirable method for clinical use, we feel that the dual-kernel algorithm which uses two region specific convolution kernels and a variable scatter fraction curve is the pref erable method.