Premium
A multiresolution restoration method for cardiac SPECT imaging
Author(s) -
Franquiz Juan M.,
Shukla Shailendra
Publication year - 1998
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.598462
Subject(s) - attenuation , imaging phantom , correction for attenuation , image restoration , spect imaging , iterative reconstruction , nuclear medicine , mathematics , computer science , physics , optics , artificial intelligence , image processing , medicine , image (mathematics)
In this study we present a multiresolution based method for restoring cardiac SPECT projections. Original projections were decomposed into a set of sub‐band frequency images by using analyzing functions localized in both the space and frequency domain. This representation allows a simple denoising and restoration procedure by discarding high‐frequency channels and performing inversion only in low frequencies. The method was evaluated in bull's eye reconstructions of a realistic cardiac chest phantom with a custom‐made liver insert and99 mTc liver‐to‐heart activity ratios (LHAR) of 0:1, 1.5:1, 2.5:1, and 3.5:1. The cardiac phantom in free air was used as the reference standard. Reconstructions were performed by filtered backprojection using (1) no correction; (2) restoration without attenuation correction; (3) attenuation correction without restoration; and (4) restoration and attenuation correction. The attenuation correction was carried out with the Chang's method for one iteration. Results were compared with those obtained using an optimized prereconstruction Metz filter. Quantitative analysis was performed by calculating the normalized chi‐square measure and mean ± s . d . of bull's eye counts. In reconstructions with high liver activity ( LHAR > 2 ) , attenuation correction without restoration severely distorted the polar maps due to the spillover of liver activity into the inferior myocardial wall. Both restoration methods when combined with an attenuation correction compensated this artifact and yielded uniform polar maps similar to that of the standard reference. There was no visual or quantitative difference between the performance of Metz filtering and multiresolution restoration. However, the main advantage of the multiresolution method is that it states a more concise and straightforward approach to the restoration problem. Multiresolution based methods does not require information about the object image or optimization processes, such as in conventional nuclear medicine restoration filters. In addition, the method is easy to implement using DFT techniques and potentially can be extended to noniterative spatially shift‐invariant restorations in SPECT.