A 3D multi resolution local analysis approach for correction of partial volume effects in emission tomography

Partial volume effects (PVE) in nuclear imaging blur contours and decrease intensity in small tissues. Among the methodologies considered for PVE correction, a new voxel-wise mutual multi-resolution approach (MMA) was recently introduced. MMA is based on the extraction and transformation of high resolution details from an anatomical image (MR or CT) and their subsequent incorporation into a low-resolution image (PET or SPECT). This voxel-wise approach allows lacking details of the low-resolution image to be reconstructed as well as contours and intensities to be recovered. However, artifacts may be introduced due to the required global linear correlation between both modalities and the current 2D implementation whereas PVE is a 3D phenomenon. The aim of the present study was to assess a new model, designed to overcome these two issues. A 3D wavelet decomposition process was built to replace the 2D global linear model with a 3D local analysis using a moving window. The algorithm was validated on synthetic and simulated images, and its performance was compared to the previous approach. The obtained quantitative correction was found to be similar to the 2D global correction in correlated cases whereas in the uncorrelated cases such as whole body or brain analysis, the new model outperforms the 2D approach without artefacts and with improved visual and quantitative correction.