Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf
|Total number to be selected: 1 Title record|
Comparison of MLAA-derived attenuation maps with and without utilisation of time-of-flight information in the attenuation estimation step
As is well known, quantitative combined PET/MR imaging depends on accurate MR-based attenuation correction (MRAC). While a mostly satisfactory state of affairs has been reached today, problems persist regarding segmentation
errors including unsatisfactory bone identification and residual systematic differences in comparison to PET/CT. Alternative or complementary strategies for attenuation correction (AC), therefore, are of considerable relevance. In this context, Maximum Likelihood reconstruction of Attenuation and Activity (MLAA) is one of the most promising approaches. As A. Rezaei et al. have shown , Time-Of-Flight (TOF) image reconstruction is required to eliminate possible ”crosstalk” between the estimated activity and attenuation distribution. On the other hand, it is widely believed that use of the TOF information during attenuation estimation does not result in image quality improvement and thus is unnecessary, see for example ref. . However, so far this assumption has never been thoroughly tested. We address this issue in the present investigation. To this end, we have compared TOF and non-TOF versions of the attenuation estimation algorithm as part of MLAA within the framework of our previously developed Tube of response High resolution OSEM Reconstruction (THOR), see ref .
MLAA is an image reconstruction algorithm, which maximizes the Likelihood function by alternately updating activity distribution and attenuation map. Maximum-Likelihood Estimation-Maximization (MLEM) is normally used for the
activity estimation and Maximum-Likelihood Transmission Reconstruction (MLTR) for the attenuation estimation. In order to investigate the potential impact of using TOF-MLTR instead of nonTOF-MLTR in the MLAA workflow both of them were implemented as a part of our THOR application. List-mode MLEM algorithm was used for activity reconstruction and accelerated by utilizing ordered subsets. For scatter correction (SC), the time-of-flight extension of the Single Scatter Simulation algorithm (SSS) was used, see ref . Attenuation map reconstruction was performed by ordered subsets accelerated list-mode version of MLTR, which is equivalent to the standard sinograms based MLTR in the non-TOF case. For the initial attenuation map estimate the MR-derived outline of the scanned object was uniformly filled with the attenuation coefficient of water. During reconstruction, attenuation map estimates were augmented by a pre-computed template of the patient bed. The main difference between TOF- and nonTOF-MLTR is the way how scatter and randoms corrections are handled. TOF information allows to individually compute this correction for each event (or TOF-bin) depending on event position along the LOR, while this correction is assumed to be the same for all the events within the LOR in the non-TOF algorithm. Consequently, any differences between both MLTR versions should be most pronounced for high contrast objects as is the case, e.g., if the bladder is within the field-of-view. Therefore, two different configurations of the whole body phantom L981602 were used. The phantom in configuration A has two cylindrical air-filled inserts and one cylindrical bone-like insert. This phantom allows to assess accuracy of the attenuation map estimate under low-contrast conditions. The phantom in configuration B comprises a large spherical ”bladder” insert with high target-to-background contrast and a small ”lesion” insert with lower contrast. The attenuation map is uniform in this case, which facilitates detection of scatter-related artifacts in the MLAA reconstructed attenuation image. Transmission scans of the phantoms with the Siemens HR+ scanner were performed and used as ground-truth for the attenuation maps.
The whole body phantom in both configurations was scanned with the Time-Of-Flight capable Philips Ingenuity-TF PET/MR scanner (TOF resolution (FWHM): ∼600 ps). Acquired data were reconstructed with THOR MLAA and TOF-MLTR and nonTOF-MLTR, respectively. In the case of configuration A (low activity contrast, high attenuation contrast) TOF-MLTR does not improve attenuation coefficients estimate significantly. Reconstructed attenuation values differ by less than 1% for bone and less than 15% for air. The situation is different for configuration B (high activity contrast, homogeneous attenuation). Due to presence of the large hot object in the field-of-view a massive artifact appears in the transaxial plane of the reconstructed attenuation map containing the ”bladder” insert. In the coronal view this artifact appears as a rather large area of apparently reduced attenuation in the middle of the phantom. The difference between the attenuation coefficient of the water background in the central and the peripheral zones depends on the reconstruction method used. Specifically, the use of TOF-MLTR instead of nonTOF-MLTR yields twofold decrease of the artifact. On the other hand, the attenuation coefficient inside the ”bladder” is about 12% higher than the reference value with TOF-MLTR compared to a 6% overestimate with nonTOF-MLTR (where this reduced deviation probably is due to the influence of the mentioned attenuation artifact).
Our preliminary results indicate that the use of TOF-MLTR within the MLAA framework provides only small improvements in terms of attenuation map accuracy if activity contrasts are modest. However, it can distinctly decrease scatter related artifacts in the presence of high activity contrast such as is frequently observed in the pelvis region. We hypothesize the advantages of TOF-MLTR will become even more apparent with increasing TOF resolution. A more detailed investigation of the benefits of TOF-MLTR usage within the MLAA workflow is under way.
 A. Rezaei, M. Defrise, G. Bal, C. Michel, M. Conti, C. Watson, and J. Nuyts, “Simultaneous reconstruction of activity and attenuation in time-of-flight PET.” IEEE transactions on Medical Imaging, vol. 31, no. 12, pp. 2224–33, dec 2012.
 A. Rezaei and J. Nuyts, “Simultaneous reconstruction of the activity image and registration of the CT image in TOF-PET,” IEEE Nuclear Science Symposium Conference Record, vol. 1852, p. 1852, 2016.
 A. Lougovski, F. Hofheinz, J. Maus, G. Schramm, E. Will, J. van den Hoff, and J. van den Hoff, “A volume of intersection approach for on-the-fly system matrix calculation in 3D PET image reconstruction,” Physics in Medicine and Biology, vol. 59, no. 3, pp. 561–577, feb 2014.
 C. C. Watson, “Extension of Single Scatter Simulation to Scatter Correction of Time-of-Flight PET,” IEEE Transactions on Nuclear Science, vol. 54, no. 5, pp. 1679–1686, 2007.
Keywords: PET; TOF-PET; PET/MR; MLAA; MRAC; Attenuation Correction
PSMR 2017 - 6th Conference on PET-MRI and SPECT-MRI, 29.-31.05.2017, Lisbon, Portugal