Mehmet Aykac

Simultaneous reconstruction of emission activity and attenuation coefficient distribution from TOF data, acquired with external transmission source

The simultaneous PET data reconstruction of emission activity and attenuation coefficient distribution is presented, where the attenuation image is constrained by exploiting an external transmission source. Data are acquired in time-of-flight (TOF) mode, allowing in principle for separation of emission and transmission data. Nevertheless, here all data are reconstructed at once, eliminating the need to trace the position of the transmission source in sinogram space. Contamination of emission data by the transmission source and vice versa is naturally modeled. Attenuated emission activity data also provide additional information about object attenuation coefficient values. The algorithm alternates between attenuation and emission activity image updates. We also proposed a method of estimation of spatial scatter distribution from the transmission source by incorporating knowledge about the expected range of attenuation map values. The reconstruction of experimental data from the Siemens mCT scanner suggests that simultaneous reconstruction improves attenuation map image quality, as compared to when data are separated. In the presented example, the attenuation map image noise was reduced and non-uniformity artifacts that occurred due to scatter estimation were suppressed. On the other hand, the use of transmission data stabilizes attenuation coefficient distribution reconstruction from TOF emission data alone. The example of improving emission images by refining a CT-based patient attenuation map is presented, revealing potential benefits of simultaneous CT and PET data reconstruction.

APD performance in light sharing PET applications

An APD-based, light-sharing detector has been evaluated for use in PET. The detector configuration is a 2 /spl times/ 2 array of 5 mm /spl times/ 5 mm Hamamatsu S8664-55 APDs used to readout multi-crystal LSO blocks. Initially, a basic performance study was undertaken with a single APD coupled to a chemically etched 4 mm /spl times/ 4 mm /spl times/ 10 mm LSO crystal (teflon wrapped) using a custom, single-channel fast ASIC preamplifier. Timing and energy resolution measurements with a /sup 22/Na source were performed using the monolithic LSO crystal coupled to the API). The timing resolution of the APD channel in coincidence with a plastic scintillator coupled to a PMT was 870 ps FWHM. The energy resolution of the 511 keV photopeak was 12.1% FWHM. Based on these initial results, a 9 /spl times/ 9 array of 2 mm /spl times/ 2 mm /spl times/ 20 mm LSO crystals was assembled and evaluated with a 2 /spl times/ 2 APD array. The LSO block had an average energy resolution of 20.9% and a timing resolution of 2.47 ns FWHM. These results show that APDs are promising photodetectors for high-resolution and cost-effective PET systems utilizing light-sharing block detectors.

Design Considerations of Phoswich Detectors for High Resolution Positron Emission Tomography

A way to improve the spatial resolution in positron emission tomography (PET) is to determine the depth-of-interaction (DOI) in the detector. A way to achieve this is to use the phoswich approach, a detector with two or more layers of different scintillators. The layer identification is done by using differences in scintillation decay time and pulse shape discrimination techniques. The advantages of the concept have been demonstrated in the HRRT high resolution PET system using a LSO/LYSO combination giving a high spatial resolution uniformity of around 2.5 mm within a larger part of the imaged volume. A phoswich combination that lately has received attention is LuAP/LSO or LuYAP/LSO. The suggestions come from the crystal clear collaboration and there is a patent application for its use in PET. This particular combination of phoswich may, however, have a complication since both LuAP and LuYAP emit in the excitation band of LSO, thus making the functionality more complex. In the present paper we have looked into this and suggested different ways to overcome potential drawbacks.

Digital timing: sampling frequency, anti-aliasing filter and signal interpolation filter dependence on timing resolution

The main focus of our study is to investigate how the performance of digital timing methods is affected by sampling rate, anti-aliasing and signal interpolation filters. We used the Nyquist sampling theorem to address some basic questions such as what will be the minimum sampling frequencies? How accurate will the signal interpolation be? How do we validate the timing measurements? The preferred sampling rate would be as low as possible, considering the high cost and power consumption of high-speed analog-to-digital converters. However, when the sampling rate is too low, due to the aliasing effect, some artifacts are produced in the timing resolution estimations; the shape of the timing profile is distorted and the FWHM values of the profile fluctuate as the source location changes. Anti-aliasing filters are required in this case to avoid the artifacts, but the timing is degraded as a result. When the sampling rate is marginally over the Nyquist rate, a proper signal interpolation is important. A sharp roll-off (higher order) filter is required to separate the baseband signal from its replicates to avoid the aliasing, but in return the computation will be higher. We demonstrated the analysis through a digital timing study using fast LSO scintillation crystals as used in time-of-flight PET scanners. From the study, we observed that there is no significant timing resolution degradation down to 1.3 Ghz sampling frequency, and the computation requirement for the signal interpolation is reasonably low. A so-called sliding test is proposed as a validation tool checking constant timing resolution behavior of a given timing pick-off method regardless of the source location change. Lastly, the performance comparison for several digital timing methods is also shown.

Simultaneous reconstruction of emission activity and attenuation coefficient distribution from TOF data, acquired with rotating external line source

The simultaneous PET data reconstruction of emission activity and attenuation coefficient distribution is under investigation, where the attenuation image is constrained by exploiting an external transmission source. Data are acquired in TOF mode, allowing in principle for separation of emission and transmission data. Nevertheless, here all data are reconstructed at once, eliminating the need to trace the position of the transmission source in sinogram space. Contamination of emission data by the transmission source and vice versa is naturally modeled. Attenuated emission activity data also provide additional information about object attenuation coefficient values. The algorithm alternates between attenuation and emission activity image updates. We also proposed a method of estimation of spatial scatter distribution from the transmission source by incorporating knowledge about the expected range of attenuation map values. The reconstruction of experimental data from the mCT scanner suggests that simultaneous reconstruction improves attenuation map image quality, as compared to when data are separated. On the other hand, the use of transmission data stabilizes attenuation coefficient distribution reconstruction from TOF emission data alone.

Performance Study of the New Hamamatsu R9779 and Photonis XP20D0 Fast 2” Photomultipliers

The focus of this paper is the evaluation of the new fast 51 mm-diameter, 8-stage Hamamatsu R9779 photomultipliers (PMTs) with an acceleration-ring at the front-end and the Photonis XP20D0 PMTs with a screening grid in front of the anode. The following performance characteristics are presented: Timing resolution, anode-scan-uniformity and transit-time spread. The unfolded timing resolution for two R9779 was 192 ps and 210 ps using plastic scintillators. The individual timing resolutions for two XP20D0 using plastic scintillators are 181 ps and 154 ps, respectively. The variation in time resolution across the windows of the two R9779 ranged between 117 ps and 171 ps, and 79 ps and 73 ps for the two XP20D0 PMTs.

CURRENT AND FUTURE USE OF LSO: CE SCINTILLATORS IN PET

Single crystal scintillators of Lu2SiO5:Ce (LSO:Ce) were first developed about 15 years ago and have been in commercial use in positron emission tomography systems for more than five years. Annual ...

Comparison of typical scintillators for PET

There has been a plethora of literature describing the various properties of scintillators that are commonly used in PET detectors. In this literature there usually is a comparison made with respect to light output and energy resolution. Unfortunately, any of these comparisons are misleading because the treatment of the different scintillators is not optimized to produce the best results that may, be possible through proper surface treatment of the scintillator. In this research, there is a comparison made between the following scintillators: BGO, LSO, GSO, LYSO, and YSO. Again, the figures-of-merit for comparison were light output and energy resolution at 511 keV.

Dynode-timing method for PET block-detectors

The focus of this paper is the investigation of a new dynode-timing technique optimized for PET block detectors. This method allows utilization of dynode signals from single but especially multiple photo-multiplier tubes (PMTs), operated with negative high-voltage. The technique will provide an event-timing trigger without deteriorating the anode signal. A printed circuit board has been developed and built for this investigation. Benchmark measurements have been performed, comparing timing of the anode signal with timing of the inverted last-dynode signal and timing of the dynode signal extracted via a newly developed LVPECL-logic based board. Timing measurements were performed with plastic as well as LSO scintillators. From single PMT measurements we find a 30 ps improvement with the dynode-timing method compared to the standard anode timing with two Photonis XP2020Q PMTs with LSO (10 mmtimes10 mmtimes10 mm). For a quad-PMT block detector, assembled of four Hamamatsu R9800 with a Hi-Rez block, the timing-resolution improves ~10%, by 43 ps compared to the standard anode timing.

Depth of Interaction With a 3-Dimensional Checkerboard Arrangement LSO-LSO Block

In order to improve image quality in Positron Emission Tomography (PET) different routes are being pursued such as fast timing resolution for time-of-flight PET, higher spatial resolution by the use of smaller scintillator pixels and the use of depth-of-interaction information. The detection of the depth-of-interaction (DOI) of a gamma ray within a detector, deploying pulse shape discrimination (PSD), has been used to increase sensitivity and spatial resolution, especially at the edge of the field of view (FOV). The DOI information is used to reduce the parallax error; thus improving spatial resolution. Commonly, different scintillator materials with different decay times and light output and other differentiating factors, such as density, emission spectra, etc., are used for DOI detectors. We present a multilayer phoswich detector comprised of LSO with different decay times in the range from 30 ns to 47 ns. The difference in decay times is achieved by co-doping LSO:Ce with Ca, resulting in short decay times of ~30 ns. The use of a cut light guide allows the use of regular Photomultiplier tubes, giving the opportunity of a potential DOI detector replacement for current detectors. We were able to identify each pixel in the different detector layers and thus, to determine the depth of the photon interaction and achieve a timing resolution of 345 ps.