Generating attenuation maps using differential attenuation data

A method is proposed to estimate attenuation from differential attenuation information (DAI) contained solely in SPECT emission data produced by ²⁰¹Tl or other multiple emission isotopes. The total attenuation along each projection line is shown to be related to the difference in attenuation for two emissions at distinct energies (low energy gamma rays are attenuated more than high energy gamma rays); this difference is reflected by the relative gamma-ray intensity detected at the two energies after compensation for Compton scatter. The feasibility of attenuation compensation using DAI was investigated using synthesized scatter-free data from an anthropomorphic digital phantom. Two approaches were investigated: (1) filtered backprojection reconstruction of sinograms pre-corrected for attenuation using a point-source approximation; and (2) incorporating the DAI into an iterative regularized image reconstruction algorithm using a more realistic distributed source model. Both these techniques improved image quality and reduced attenuation artifacts visible in uncorrected filtered backprojection images. However, Poisson noise presents a significant obstacle to practical application of this technique and several enhancements are required before it can be successfully applied to realistic imaging data