Compensation for nonuniform attenuation in SPECT brain imaging

Accurate compensation for photon attenuation is needed to perform quantitative brain single-photon-emission computed tomographic (SPECT) imaging. Bellini's attenuation-compensation method (1979) has been used with nonuniform attenuation map to account for the nonuniform attenuation properties of the head. Simulation studies using a three-dimensional (3-D) digitized anthropomorphic brain phantom were conducted to compare quantitative accuracy of reconstructions obtained with the nonuniform Bellini method to that obtained with the Chang method (1978) and to iterative reconstruction using maximum-likelihood expectation maximization (ML-EM). Using the Chang method and assuming the head to be a uniform attenuator gave reconstructions with an average bias of approximately 6-8%, whereas using the Bellini or the iterative ML-EM method with a nonuniform attenuation map gave an average bias of approximately 1%. The computation time required to implement nonuniform attenuation compensation with the Bellini algorithm is approximately equivalent to the time required to perform one iteration of ML-EM. Thus, using the Bellini method with a nonuniform attenuation map provides accurate compensation for photon attenuation within the head, and the method can be implemented in computation times suitable for routine clinical use.