TY - JOUR T1 - Diaphragm tracking using shape modelling for motion-correction in cardiac PET JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 588 LP - 588 VL - 50 IS - supplement 2 AU - Sarah McQuaid AU - Tryphon Lambrou AU - Vincent Cunningham AU - Valentino Bettinardi AU - Maria Gilardi AU - Brian Hutton Y1 - 2009/05/01 UR - http://jnm.snmjournals.org/content/50/supplement_2/588.abstract N2 - 588 Objectives Diaphragm motion should be considered when motion-correcting cardiac studies, since it can cause attenuation-correction artefacts via PET-CT mismatch. We propose to track the diaphragm in gated PET data with the aid of a single CT and a Statistical Shape Model (SSM), which is trained using segmented diaphragm surfaces. This move towards a data-driven approach eliminates the need for a gated CT, thereby lowering dose and reducing the problem of mismatch. Methods A respiratory-gated FDG PET-CT patient study was used for model testing. Points on the diaphragm were identified manually in the PET frames, which were sparse due to the low contrast. The diaphragm was also fully segmented in the CT for validation purposes. The model was fitted to: a) the PET-derived points, b) the PET-derived points with one frame replaced by the equivalent CT frame, to mimic a single CT and c) the CT segmented surfaces. Results The residual errors per point with 5 fitted modes in each case were: a) 4.0, b) 3.5 and c) 3.4 mm. Combining a single CT frame with PET had a lower error than PET only, indicating that the CT provided valuable shape information. However replacing the full CT surfaces with the combined PET-CT points had only a small effect on model-fitting accuracy, showing the robustness of the model to missing and noisy data. Conclusions The results demonstrate the suitability of a SSM for diaphragm tracking with gated PET and a single CT. This could therefore provide an alternative to a gated CT acquisition and could overcome many of the associated practical issues. ER -