The influence of attenuation correction and reconstruction techniques on the detection of hypo-perfused lesions in brain SPECT images

Abstract

We evaluated the effects of attenuation correction and reconstruction techniques on the detection of hypoperfused lesions in brain SPECT imaging. A software phantom was constructed using the data available on the BrainWeb database by assigning activity values to grey and white matter. The true attenuation map was generated by assigning attenuation coefficients to six different tissue classes to create a non-uniform attenuation map. The uniform attenuation map was calculated using an attenuation coefficient of 0.15 cm(-1). Hypoperfused lesions of varying intensities and sizes were added. The phantom was then projected as typical SPECT projection data, taking into account attenuation and collimator blurring with the addition of Poisson noise. The projection data were reconstructed using four different methods: filtered back-projection in combination with Chang's first-order attenuation correction using the uniform or the true attenuation map and maximum likelihood iterative reconstruction using the uniform or the true attenuation map. Different Gaussian post-smoothing kernels were applied onto the reconstructed images and the performance of each procedure was analysed using figures of merit such as signal-to-noise ratio, bias and variance. Uniform attenuation correction offered only slight deterioration of the signal-to-noise ratio compared to the true attenuation map. Maximum likelihood produced superior signal-to-noise ratios and lower bias at the same variance in comparison to the filtered back-projection. Uniform attenuation correction is adequate for lesion detection while maximum likelihood provides enhanced lesion detection when compared to filtered back-projection.