Correction of photon attenuation and collimator response for a body-contouring SPECT/CT imaging system.

UNLABELLED (111)In-Capromab pendetide imaging is indicated for postprostatectomy patients at risk for residual or recurrent disease. However, this study is complicated by relatively long times for tumor uptake and background washout that require imaging to be performed several days after radiopharmaceutical administration. In addition, (111)In-capromab pendetide demonstrates uptake in normal structures that produce images that are interpreted best using correlation with anatomic imaging. Finally, the visual quality of radionuclide imaging can be improved with corrections for photon attenuation and for the geometric response of the radionuclide collimator. Therefore, we have evaluated the advantages of using a commercially available dual-modality SPECT/CT system. In this article, we evaluate a novel iterative reconstruction algorithm using the SPECT/CT data obtained from phantoms and (111)In-capromab pendetide patient studies. METHODS Phantom data acquired with the dual-head SPECT camera were reconstructed using both filtered backprojection (FBP) and an iterative maximum-likelihood expectation maximization (MLEM) algorithm incorporating corrections for (a) attenuation coefficient at the effective energy of the radionuclide (either (99m)Tc or (111)In) and (b) collimator response based on experimentally measured depth-dependent spatial resolution of the camera. The collimator response model used the coregistered CT image to estimate the source-target distances produced by the patient-contouring logic of the SPECT camera. Spatial resolution was measured using SPECT images of 2 line sources and uniformity from a uniform cylindric tank. Clinical (111)In-capromab pendetide SPECT/CT data were acquired according to the radiopharmaceutical manufacturer's protocol. Region-of-interest (ROI) analysis of a transverse slice at the level of the sacral base produced mean, median, maximum, and minimum counts per pixel for bone marrow and surrounding soft-tissue ROIs. Ratios of the mean capromab pendetide uptake within marrow to uptake within soft tissue were compared for images reconstructed with FBP versus that obtained from the MLEM method with photon attenuation and collimator response corrections. RESULTS The source-target distances reconstructed from the patient-specific CT image agreed well with the corresponding values recorded manually from the camera display unit. This information was incorporated into the iterative reconstruction algorithms and improved the quality of SPECT images from phantoms and patients versus SPECT images reconstructed without the depth-dependent collimator response model. Qualitatively, SPECT images reconstructed with corrections for photon attenuation and collimator response showed less background activity and improved target contrast compared with those images reconstructed with FBP. The target-to-background ratio (marrow uptake-to-soft-tissue uptake) was significantly better using MLEM reconstruction than with FBP when mean uptake values were measured. CONCLUSION A priori anatomic data can be used to enhance the quality of the SPECT image when reconstructed using iterative techniques (e.g., MLEM) that use the CT data to produce a patient-specific attenuation map and a collimator response model based on the body contour produced during the SPECT acquisition.