RT Journal Article SR Electronic T1 Estimating an attenuation map from measured scatter for 180o cardiac SPECT JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 1357 OP 1357 VO 51 IS supplement 2 A1 Sarah Cade A1 Alexandre Bousse A1 Simon Arridge A1 Martyn Evans A1 Brian Hutton YR 2010 UL http://jnm.snmjournals.org/content/51/supplement_2/1357.abstract AB 1357 Objectives Attenuation correction is required for accurate reconstruction of cardiac SPECT images, however inclusion of an accurate attenuation map is essential. This work aims to establish a method for estimating attenuation from measured scatter data. Methods A photopeak-based reconstruction algorithm exists which estimates both the attenuation map (AM) and activity from just photopeak emission data (Nuyts et al. IEEE TMI, 18:393-403, 1999). The algorithm performs alternating iterations of MLEM to estimate activity and a gradient ascent method to estimate AM. A similar alternating algorithm has been developed which instead uses measured scatter data and a model of the relationship between scatter and attenuation to estimate the AM. The scatter-based algorithm has been tested using the XCAT phantom in an idealised situation (i.e. no noise, no photopeak scatter, perfect resolution) and compared to the photopeak-based method. Monte Carlo (MC) simulations have also been used to assess the scatter-based algorithm. Results Photopeak-based reconstruction using 360o data produced an AM with accurate body and lung outlines but significant cross-talk between emission and attenuation data. Using 180o data, however, lung outlines could not be identified. In contrast the scatter-based algorithm demonstrated accurate reconstruction for both 360o and 180o ideal data. The algorithm showed promise using MC data although some cross-talk issues occur that may necessitate inclusion of segmentation. The mean square errors for estimated AMs compared to the true XCAT AM were: Conclusions The proposed scatter-based algorithm offers a promising approach for AM reconstruction without a transmission scan. Work is in progress to further optimise the scatter model and to account for multiple scatter