Time-of-flight in cardiac PET/TC: What do we know and what we should know?
Time-of-flight in cardiac PET/TC: What do we know and what we should know?
Roberta Matheoud 0 2
Michela Lecchi 0
0 Reprint requests: Roberta Matheoud, PhD, Department of Medical Physics, University Hospital Maggiore della Carita` , C.so Mazzini, 18, 28100 Novara , Italy
1 Health Physics, San Paolo Hospital, University of Milan , Milan , Italy
2 Department of Medical Physics, University Hospital Maggiore della Carita` , Novara , Italy
Coronary arteries disease (CAD) is the most
common cause of death in the world, and an accurate
diagnosis can be made by combining the diagnostic
results of different invasive and noninvasive techniques.
Among the latter, molecular imaging has gained
growing importance, and above all, positron emission
tomography (PET), combined with computed
tomography (CT), has shown increasing diagnostic accuracy in
detecting small and low perfusion areas in the
myocardial wall.1 This was possible thanks to the
cardiacdedicated PET radiotracers (for example, Rubidium-82
chloride, 82Rb), but also to the last generation of
correction techniques available for the acquisition of the
emission data and in the reconstruction process of
Time-of-flight (TOF) technology allows for the
identification of positron annihilation location with
greater accuracy than non-TOF detection, resulting in a
reduction of the spatial uncertainty and thus, in lowering
the image noise. Accordingly, an advantage of TOF is
the improvement of lesion contrast and the related
signal-to-noise ratio, finally resulting in a superior overall
Effects on PET images related to the finite spatial
resolution of the PET scanners are referred to as partial
volume effect (PVE) and result in a quantitative bias
when small objects are investigated.3 The
PVE-correction algorithms available on the last generation of PET
systems belong to the resolution modeling (RM)
technique that incorporates the point spread function (PSF)
of the system used in the forward and backward
projection matrix of the iterative process. However,
although the application of PSF modeling has
demonstrated to improve image quality of Fluorine-18
fluorodeoxyglucose ([18F]FDG) studies, there is no
consensus on the actual benefits derived on quantitation,
as the noise propagation related to RM is far from being
Several papers have studied the impact of TOF
technology and PSF modeling on the detectability of hot
lesions in oncological imaging with [18F]FDG,5 but their
effects in the presence of cold lesions or low-uptake
regions, such as those of myocardial imaging, are far
from being fully explored. Moreover, the study of
perfusion (and/or viability) of the myocardium is further
complicated by the organ motion and by the
heterogeneity of radiotracer distribution and tissue density in
the human torso.
A further step toward the assessment of these issues
in PET cardiac studies was recently achieved by Dasari
et al.6 In a group of obese patients, the authors studied
the individual and combined effects of TOF and PSF on
82Rb myocardial perfusion images as a function of
different anatomic features: gender, body mass index
(BMI), cross-sectional body area in the scanner field of
view, and left-ventricular myocardial volume.
In accordance with a previous published study,7
TOF confirmed the gain in uniformity of the radiotracer
distribution, showing in particular an increased uptake
of about 10% in the septal segments of the myocardial
wall. However, the element of originality of the study of
Dasari et al is that the increased uptake at septal level is
more pronounced for patients with large cross-sectional
area and for female (due to breast attenuation). On the
contrary, PSF modeling had only a slight effect on
image quality, increasing the overall perfusion uptake
irrespective of patient gender and anatomic features.
The perfusion semiquantitative analysis reported in
this study showed somehow surprising results: the
summed stress (SSS) and rest (SRS) scores were, on
average, significantly smaller for TOF with respect to
non-TOF reconstructions. These results would indicate a
reduction in the appearance of the defects in TOF
images that could affect the risk stratification of the patients
with regard to cardiac event-free survival. Consequently,
some patients with a severe disease after a non-TOF
PET study could be moved to a less-severe category if
TOF reconstruction is used.
Thus, does TOF improve the uniformity of the PET
cardiac images, and, at the same time, reduce the
appearance of the perfusion defects particularly in obese
How much benefit might be derived from TOF
technology in the myocardial perfusion studies is still
unclear and needs to be better understood.
TOF AND PSF VERSUS CONVERGENCE
OF ITERATIVE ALGORITHM IN CARDIAC
TOF demonstrated its superiority in the contrast
recovery of hot lesion and in the better visualization of
cold areas in studies performed on both
anthropomorphic phantoms and on [18F]FDG patients.4 The evidence
was that TOF provides overall significant improvement
in image uniformity and increased convergence rate of
the iterative algorithm compared to non-TOF
technology, which requires approximately three times more
iterations than TOF to reach the true activity distribution
in the reconstructed images. This issue is of particular
importance in nuclear cardiology, as the basic task is to
determine the presence or the absence of ipo-perfused
regions in the myocardial wall: for a normal patient, a
uniform tracer distribution in the myocardial wall has to
be translated into a set of uniform images of the
leftventricular signal. However, the convergence is a
nonlinear process and behaves differently depending
on the different radiotracer concentrations present in the
field of view, as areas with lower uptake (such as the
perfusion defects) converge slower than those with
higher uptake.8 Heart orientation, breast size and
location, and the presence of fluid in lung could also affect
the image uniformity due to the incomplete convergence
of iterative algorithm in the corresponding areas of the
The synergic effect of PSF when coupled to TOF
has been reported in the literature for both [18F]FDG
) and phantom (
). The authors agree in
the contrast improvement of cold defects, although at a
different level on different PET/CT scanners.9
The PSF modeling is obtained by an analytical
approach,10 via Monte Carlo simulations11 or by
measured datasets,12 depending on the model of PET
scanner. Anyway, in contrast with TOF, PSF model
included in the iterative algorithm degrades the
performances slowing the convergence of the iterative
Moreover, the study of Dasari et al used an
approximate PSF modeling as the PSF of 18F was used
for 82Rb exams. The effect of PSF modeling on 82Rb
imaging could be in principle different from that on 18F,
due to the higher energy of the positron emitted by 82Rb
(1535 keV) with respect to 18F (635 keV). This could
explain the small increase in perfusion uptake in almost
all heart segments (averaging 1.5%), with no statistical
significance for any segment when PSF modeling is
Few other papers studied the effect of PSF on 82Rb
imaging, but they are all related to blood flow and
myocardial flow reserves in patients.13–15 Anyway, they
all show equivalency or superiority of TOF-only PET
TOF AND PET/TC MISALIGNMENT ARTIFACTS
Misalignment between PET and CT images is
known to generate artifactual defects in PET images
due to inaccurate attenuation correction (AC). This is
more critical in cardiac studies where, in addition to the
patient movements, heart beat and respiratory motion
can produce an involuntary spatial mismatch between
PET and CT. This could result in an unrealistic
increased or decreased cardiac uptake corresponding to
over- or under-attenuation correction of the emission
It was observed in a phantom study that the
mismatch between PET and CT is more evident for
non-TOF than TOF images. TOF reconstruction is less
sensitive to mismatched attenuation correction,
erroneous normalization, and poorly estimated scatter
correction, as TOF provides additional information of
the origin of each detected event. The mismatched
artifacts can be also observed in patients, but with a
Dasari et al suggest indeed that the increase in
septal wall perfusion when TOF is used could be
attributable to inconsistencies in scatter correction that
can be more critical for areas that are located centrally in
the body, as is the case in respect of the septal wall.
The justification for the superiority of TOF images
lies in the fact that in the iterative algorithm, the
quantitative contribution coming from TOF has superior
weight to that coming from the AC based on CT, when
the mismatch between PET and CT images is small
(quantitation errors \ 10% for 5 mm misalignment).16
When PET/CT misalignment is greater than 5 mm, the
use of reconstruction algorithms—different from the
traditional ordered subsets expectation maximization
(OSEM) used by Dasari et al—has been proposed. The
maximum likelihood attenuation and activity (MLAA)
algorithm in the paper of Presotto et al17 is a feasible and
robust technique to avoid large mismatch artifacts in
TOF PET cardiac studies, provided that a CT of the
patient is available.
TOF AND OVERWEIGHT PATIENTS
The larger the cross-sectional area present in the
field of view, the stronger the attenuation and scatter
phenomena of the emission data and the higher the
probability of mismatch between PET and CT images.
The category of obese patients with CAD would clearly
benefit from the above-described potentialities of TOF.
Moreover, with TOF, the gain of signal-to-noise ratio
generally increases with the increasing patient size.
82Rb is widely used for PET myocardial perfusion
study thanks to its high sensitivity,18 and its distribution
in the myocardial wall of obese patients was recently
investigated by comparing TOF and non-TOF images
reconstructed with a variable number of iterations.7 It
has been demonstrated that nonconvergence (insufficient
number of iterations) of non-TOF reconstruction is very
important in obese patients and may result in artifactual
nonuniformity distribution of the perfusion radiotracer
with consequent both false-positive and false-negative
interpretation. This was more important for the
myocardial regions located near the lungs and the breast, where
the tissue density is far from being uniform.
TOF AND PERFUSION QUANTIFICATION
Surprisingly, the results of Dasari et al would
indicate a reduction in the appearance of the perfusion
defects in TOF-PET imaging with significant decreases
between 2.4 and 3.0 for both SRS and SSS. This could
affect risk stratification for CAD, moving some of
patients to a less-severe category. On the contrary, SRS
and SSS were largely unchanged when PSF modeling
In line with the above considerations, a possible
explanation is related to the incomplete convergence of
non-TOF reconstruction in the study of Dasari, so one
may wonder how the perfusion would be modified by
increasing the numbers of iterations. In fact, all the
iterative reconstruction types (non-TOF, TOF, and
TOF ? PSF) were performed with four iterations, 21
subsets, and an 8-mm full-width at half-maximum
(FWHM) 3D Gaussian post-reconstruction filter.
Moreover, SSS and SRS are based on the evaluation
of the differences between the patient-specific
distribution of myocardial perfusion and the corresponding
mean in normal patients as a function of normal
variability. Both the mean and variability of normal
distribution uptake depend on a number of factors,
including the reconstruction algorithm used and the
correction techniques applied. When a nonspecific
normal database is used, misestimation in the evaluation
of perfusion scores is reported in the literature.19,20
Thus, the quantitative approach is a robust tool for
patient-risk assessment, provided that specific normal
database are used. Dasari et al did not mention which
normal databases were used in the study.
A VIEW TO THE FUTURE
All the observations we reported claim that further
knowledge ought to be gained on the correct use of TOF
and non-TOF technologies in PET studies of myocardial
perfusion. New experiments under controlled conditions
with known clinical distributions of 82Rb, i.e., with
appropriate anthropomorphic phantoms with different
anatomic features, should be performed with a to better
understand the potential false-positive or false-negative
interpretation when convergence of the iterative
reconstruction is not reached. Moreover, further clinical
studies should be performed in order to evaluate the
effects of TOF and PSF on SSS and SRS results with
respect to a gold standard (e.g., coronary angiography)
rather than those of the non-TOF iterative
In the case of a clinical trial of a promising new
PET radiotracer of perfusion study, the use of the PET
correction techniques, such as TOF and PSF, should be
optimized for cardiac studies on each single PET
scanner before starting the acquisition of the enrolled
R. Matheoud and M. Lecchi have nothing to disclose in
relation to this Editorial.
1. Le Meunier L , Slomka PJ , Dey D , Ramesh A , Thomson LE , Hayes SW , Friedman JD , Cheng V , Germano G , Berman DS . Enhanced definition PET for cardiac imaging . J Nucl Cardiol 2010 ; 17 : 414 - 26 .
2. Conti M. Why is TOF PET reconstruction a more robust method in the presence of inconsistent data? Phys Med Biol 2011 ; 56 : 155 - 68 .
3. Erlandsson K , Buvat I , Pretorius PH , Thomas BA , Hutton BF . A review of partial volume correction techniques for emission tomography and their applications in neurology, cardiology and oncology . Phys Med Biol 2012 ; 57 : R119 - 59 .
4. Rahmim A , Tang J . Noise propagation in resolution modeled PET imaging and its impact on detectability . Phys Med Biol 2013 ; 58 : 6945 - 68 .
5. El Fakhri G , Surti S , Trott CM , Scheuermann J , Karp JS . Improvement in lesion detection with whole-body oncologic timeof-flight PET . J Nucl Med 2011 ; 52 : 347 - 53 .
6. Dasari PKR , Jones JP , Casey ME , Liang Y , Dilsizian V , Smith MF . The effect of time-of-flight and point spread function modeling on 82Rb myocardial perfusion imaging of obese patients . J Nucl Cardiol 2018 . https://doi.org/10.1118/1.4875725.
7. DiFilippo FP , Brunken RC . Impact of time-of-flight reconstruction on cardiac PET images of obese patients . Clin Nucl Med 2017 ; 42 ( 2 ): e103 - 8 .
8. Schaefferkoetter J , Ouyang J , Rakvongthai Y , Nappi C , El Fakhri G. Effect of time-of-flight and point spread function modelling on detectability of myocardial defects in PET . Med Phys 2014 . https://doi.org/10.1118/1.4875725.
9. Matheoud R , Lecchi M , Lizio D , Scabbio C , Marcassa C , Leva L , Del Sole A , Rodella C , Indovina L , Bracco C , Brambilla M , Zoccarato O . Comparative analysis of iterative reconstruction algorithms with resolution recovery and time of flight modeling for 18F-FDG cardiac PET: A multi-center phantom study . J Nucl Cardiol 2017 ; 24 : 1036 - 45 .
10. Lecomte R , Schmitt D , Lamoureux G . Geometry study of a high resolution PET detection system using small detectors . IEEE Trans Nucl Sci 1984 ; 31 : 556 - 61 .
11. Alessio AM , Kinahan PE , Lewellen TK . Modeling and incorporation of system response functions in 3-D whole body PET . IEEE Trans Med Imaging 2006 ; 25 : 828 - 37 .
12. Panin VY , Kehren F , Michel C , Casey M. Fully 3-D PET reconstruction with system matrix derived from point source measurements . IEEE Trans Med Imaging 2006 ; 25 : 907 - 21 .
13. Armstrong IS , Tonge CM , Arumugam P . Impact of point spread function modelling and time-of-flight on myocardial blood flow and myocardial flow reserve measurements for 82Rb cardiac PET . J Nucl Cardiol 2014 ; 21 : 467 - 74 .
14. Germino M , Ropchan J , Mulnix T , Fontaine T , Nabulsi N , Ackah E , Feringa H , Sinusas AJ , Liu C , Carson RE . Quantification of myocardial blood flow with 82Rb: Validation with 15O-water using time-of-flight and point-spread-function modeling . EJNMMI Res 2016 ; 6 : 68 - 79 .
15. Chilra P , Gnesin S , Allenbach G , Monteiro M , Prior JO , Vieira L , Pires Jorge JA. Cardiac PET / CT with 82Rb: Optimization of image acquisition and reconstruction parameters . EJNMMI Phys 2017 ; 4 : 10 - 25 .
16. Bai C , Andreyev A , Hu Z , Maniawski P , Zhang J , Knopp M. TOF reconstruction for improved quantitative accuracy in PET/CT studies with misalignment . J Nucl Med 2016 ; 57 : 478 .
17. Presotto L , Busnardo E , Perani D , Gianolli L , Gilardi MC , Bettinardi V . Simultaneous reconstruction of attenuation and activity in cardiac PET can remove CT misalignment artifacts ? J Nucl Cardiol 2016 ; 23 : 1086 - 97 .
18. Sampson UK , Dorbala S , Limaye A , Kwong R , Di Carli MF. Diagnostic accuracy of 82Rb myocardial perfusion imaging with hybrid positron emission tomography/computed tomography in the detection of coronary artery disease . J Am Coll Cardiol 2007 ; 49 ( 10 ): 1052 - 8 .
19. Zoccarato O , Marcassa C , Lizio D , Leva L , Lucignani G , Savi A , et al. Differences in polar-map patterns using the novel technologies for myocardial perfusion imaging . J Nucl Cardiol 2016 ; 24 : 1626 - 36 .
20. Scabbio C , Zoccarato O , Malaspina S , Lucignani G , Del Sole A , Lecchi M. Impact of non-specific normal databases on perfusion quantification of low-dose myocardial SPECT studies . J Nucl Cardiol 2017 . https://doi.org/10.1007/s12350-017-1079-5.