 Original research
 Open Access
 Published:
Whole tumor kinetics analysis of ^{18}Ffluoromisonidazole dynamic PET scans of nonsmall cell lung cancer patients, and correlations with perfusion CT blood flow
EJNMMI Research volume 8, Article number: 73 (2018)
Abstract
Background
To determine the relative abilities of compartment models to describe timecourses of 18Ffluoromisonidazole (FMISO) tumor uptake in patients with advanced stage nonsmall cell lung cancer (NSCLC) imaged using dynamic positron emission tomography (dPET), and study correlations between values of the blood flowrelated parameter K_{1} obtained from fits of the models and an independent blood flow measure obtained from perfusion CT (pCT).
NSCLC patients had a 45min dynamic FMISO PET/CT scan followed by two static PET/CT acquisitions at 2 and 4h postinjection. Perfusion CT scanning was then performed consisting of a 45s cine CT.
Reversible and irreversible two, three and fourtissue compartment models were fitted to 30 timeactivitycurves (TACs) obtained for 15 whole tumor structures in 9 patients, each imaged twice. Descriptions of the TACs provided by the models were compared using the Akaike and Bayesian information criteria (AIC and BIC) and leaveoneout crossvalidation. The precision with which fitted model parameters estimated groundtruth uptake kinetics was determined using statistical simulation techniques. Blood flow from pCT was correlated with K_{1} from PET kinetic models in addition to FMISO uptake levels.
Results
An irreversible threetissue compartment model provided the best description of whole tumor FMISO uptake timecourses according to AIC, BIC, and crossvalidation scores totaled across the TACs. The simulation study indicated that this model also provided more precise estimates of FMISO uptake kinetics than other two and threetissue models.
The K_{1} values obtained from fits of the irreversible threetissue model correlated strongly with independent blood flow measurements obtained from pCT (Pearson r coefficient = 0.81). The correlation from the irreversible threetissue model (r = 0.81) was stronger than that from than K_{1} values obtained from fits of a twotissue compartment model (r = 0.68), or FMISO uptake levels in static images taken at timepoints from tracer injection through to 4 h later (maximum at 2 min, r = 0.70).
Conclusions
Timecourses of whole tumor FMISO uptake by advanced stage NSCLC are described best by an irreversible threetissue compartment model. The K_{1} values obtained from fits of the irreversible threetissue model correlated strongly with independent blood flow measurements obtained from perfusion CT (r = 0.81).
Background
The radiotracer ^{18}Ffluoromisonidazole (FMISO) diffuses passively into cells, where it is reduced and irreversibly bound in hypoxic environments. Thus, positron emission tomography (PET) imaging of FMISO uptake can be used to localize hypoxic tumor subvolumes [1,2,3]. The degree of hypoxia can be estimated either from uptake levels seen in single FMISO images collected 2–4 h after tracer injection [4], or from analysis of the kinetics of FMISO uptake in dynamic sequences of PET images (dPET). The kinetic analysis provides fitted values of model rateconstants related to blood flow and FMISO transport and intracellular binding and can be performed at the whole tumor level or voxelbyvoxel. For headandneck cancers, it has generated indices that correlate with radiotherapy (RT) outcomes [5].
Blood flow is often imaged using a perfusion CT (pCT) technique first proposed in 1980, in which iodine containing contrast is injected as a bolus through a venous cannula, and its passage through the patient is dynamically imaged and kinetically analyzed on the assumption that iodine concentrations within tissues are linearly proportional to changes in measured CT attenuation [6]. Blood flow measures obtained from pCT have been found to correlate with perfusion measurements obtained from dPET imaging of ^{15}Owater uptake [7], which in turn were correlated with kinetics indices obtained from compartment modeling of the first 2 min of ^{18}Ffluorodeoxyglucose (FDG) dPET scans [8].
In this study, we investigate which of several compartment models best describes PETimaged timecourses of FMISO uptake in whole NSCLC tumors, and we identify the model whose fits to the timecourse data provide the most precise estimates of tracer kinetics rateconstants according to statistical simulations. Correlations are determined between perfusion measures obtained directly from the PET FMISO kinetics model fits and independently from pCT.
Methods
PET and pCT image acquisition and processing
In a preclinical study, the investigational drug buparlisib (Novartis) reduced tumor hypoxia in vivo [9]. A clinical trial (BKM120) completed in Oxford (NCT02128724) has the primary aim of determining the maximum tolerated dose of buparlisib in nonsmall cell lung cancer (NSCLC) patients treated palliatively using radiotherapy, and the secondary goal of validating the preclinical results in these patients, who are imaged using FMISO PET at baseline and 7 days after administration of buparlisib without any other intervention. The study has been approved by the local ethics committee and signed informed consent obtained from all patients.
Patients were imaged supine with their arms by their side using a GE Discovery 690 PET/CT scanner (GE Healthcare). They were injected with 370 MBq FMISO 30 s into PET imaging, which continued for 45 min and resumed for 10 min intervals at 2 and 4h postinjection. Prior to each PET acquisition, a CT scan was performed for localization and PET attenuation correction. PET images were reconstructed using a timeofflight ordered subset expectation maximization algorithm (VPFX, GE Healthcare). The first 45 min of data were binned into two parallel time sequences, S1 (1 s × 30 s, 12 s × 5 s, 6 s × 10 s, 5 s × 30 s, 10 s × 60 s, 6 s × 300 s) and S2 (1 s × 30 s, 60 s × 1 s, 12 s × 10 s, 3 s × 30 s, 10 s × 60 s, 6 s × 300 s), and reconstructed as images on a matrix of 5.5 mm^{3} × 5.5 mm^{3} × 3.3 mm^{3} voxels. Data collected during the two later 10min intervals were processed as single frames [10].
pCT scanning was performed immediately after PET/CT imaging concluded at 4h postinjection of FMISO, with patients set up in the same position on the same PET/CT scanner. Initially a precontrast CT scan (helical mode, 120 kV, smart mA, 32 noise index) was carried out to determine the region over which the pCT data would be collected. Then pCT scanning commenced (120 kV, 60 mA), collecting one 3D image in each of 45 consecutive seconds, over an axial length of 40 mm corresponding to the CT detector width. During pCT scanning 70 mL contrast (Omnipaque 300) was injected at 5 mL/s, followed by 25 mL water at 5 mL/s, with patients instructed to hold their breath at inspiration for as long as possible, breathing out very slowly if necessary.
For each patient, the primary tumor, involved nodes, and metastases were outlined on the PET/CT images by an experienced radiologist, and a blood region was defined within the central part of the descending aorta on five or more consecutive PET axial slices [11]. These were outlined on the CT images with the patient’s prior contrastenhanced CT imaging used to assist in determining tumor regions. Timeactivity curves (TACs) representing timecourses of mean FMISO tracer activity concentrations within each tumor volumeofinterest (VOI) and the blood region were obtained from PET sequences S1 and S2 respectively. Activity data from the 10min frames collected at 2 and 4 h postinjection were appended to the TACs. A total of 30 wholevolume tumor FMISO uptake TACs, obtained for 15 volumesofinterest (9 primary tumors, 5 involved nodes, and 1 metastasis) in 9 patients, each imaged twice were studied (Table 1). Using a standard tumor to blood ratio of 1.4 (on the static images four hours postinjection), all of these whole tumor volumes had a number of voxels within them which could be considered to be hypoxic [4]. No direct oxygen measurements of the tumors were made in this trial.
Using Hermes Hybrid Viewer software (Hermes Medical Solutions AB, Sweden), the CT images obtained at PET/CT were rigidly registered with the CT images collected just before pCT, allowing outlines of the primary tumors defined on the PET/CT to be transferred to the pCT scans. An example of a FMISO PET/CT image 4 h postinjection is shown in Fig. 1.
PET kinetics analysis and model fitting
Several methods have been used to analyze dPET data, the most common being compartment modeling [12]. Figure 2 illustrates reversible two, three, and fourtissue linear compartment models which we have fitted to timecourses of tumor tracer uptake [13,14,15]. FMISO binding is generally considered irreversible, and therefore, alongside reversible models, we have also studied irreversible models in which the rateconstant describing movement of bound to unbound tracer is set to zero. The compartment model developed by Casciari et al. [2] attempts to reflect the chemical processes that occur for FMISO uptake; however, this model has many fitting parameters, and so in this work, simpler compartment models have been investigated. We denote by xCyK a model comprising a linear chain of xtissue compartments (excluding bloodborne tracer) and y rateconstants, and in order to associate rateconstants with particular models, we add the subscript xC to the names of rateconstants, except for those of the threetissue compartment model.
Kinetics analysis was carried out using the PMOD system (PMOD Technologies) as described by McGowan et al. [10] In brief, an imagederived FMISO input function [16, 17] was obtained from the blood region outlined within the descending aorta, and a compartment model was fitted to the tumor TAC and input function data, generating fitted values of model rateconstants. Fitting was carried out by minimizing the weighted sum of squares between modeled and measured tumor uptake, using the Levenberg–Marquardt algorithm with weighting function
where t_{i} and ∆t_{i} are the midtime postinjection and duration of the ith of T frames, λ is the decay constant for ^{18}F, and C_{PET} the measured PET activity concentration at time t_{i} [15, 17].
Model fitting was initiated from 100 randomly generated sets of starting values (suitably constrained), to attempt to reach global rather than local best fits [18]. Fluxconstants were calculated from irreversible twotissue compartment model (2C3K) fits as
and from irreversible threetissue (3C5K) compartment model fits as
Assessment of PET kinetic model fits
The Wald–Wolfowitz runstest was used to determine the adequacy of descriptions of FMISO uptake TACs provided by compartment model fits [19, 20]. To further assess the relative abilities of the different models to describe the data, we used the Akaike information criterion (AIC) [21] corrected for small sample size [22], the Bayesian information criterion (BIC) [22], and leaveoneout crossvalidation [23]. The crossvalidation approach proceeded by fitting each model to the complete dataset minus one point, calculating the differences between the value of the omitted datapoint and values predicted by the models, repeating this process sequentially T times leaving out a different datapoint each time, and finally calculating the mean of the squared error of prediction (MSEP) for each model, the model with the lowest MSEP being considered best. These model selection methods use slightly different criteria to determine the model that best describes the data (all penalizing highly parameterised models which are likely to over fit the data), for completeness all have been included in this work.
A statistical simulation procedure was used to assess which model produced the most accurate and precise rateconstant estimates [10]. For this analysis, we used 3C5K and 3C6K model fits to the 30 measured whole tumor TACs to create noisefree “groundtruth” TACs binned into the same framelengths as the original data, and the parameter values of these fits were taken as groundtruth rateconstants. For each of the resulting 60 groundtruth TACs, 1000 noisy TACs were simulated by adding normally distributed random variables to the activity concentrations of the individual timeframes, the variances of the noise differing between frames according to the inverse of Eq. (1) and scaled to match noiselevels seen on the measured whole tumor TACs [10, 15, 24,25,26]. The average whole tumor scaling factor with the weighting factor used here was 0.6 ± 0.3 (one standard deviation). The simulated TACs were then fitted using the 2C3K, 2C4K, 3C5K, and 3C6K models.
The simulated noise introduces random uncertainties and systematic error (bias) into fitted parameter values, adding to any underlying bias that results from mismatches between the fitted models and the groundtruth models used to generate the simulated TACs. As described by McGowan et al. [10], for each of the 30 groundtruth TACs associated with each groundtruth model, individual biases in parameter values were determined from fits to the 1000 noise realizations, and then these bias estimates were combined to calculate the overall mean bias (MB) and variance of bias values (\( {\sigma}_B^2 \)) across the groundtruth TACs. The mean variance (\( {\sigma}_P^2 \)) was calculated for each parameter as the average of the parameter variances obtained for each of the 30 groundtruth TACs. Then, the σ_{B} and σ_{P} terms were combined to generate a total uncertainty, σ_{T}, given by
For some fitted models, certain individual rateconstants are not uniquely related to any single groundtruth model parameter: for example, the processes described by the K_{1–2C} parameter of twocompartment models are split between rateconstants K_{1} and k_{3} in threecompartment models. For such rateconstants, therefore, only the σ_{P} values were calculated.
Perfusion CT analysis and comparison with PET kinetic modeling
The small size of pCT image voxels (0.7 mm ×0.7 mm × 50 mm) makes parametric images of perfusion susceptible to movement [27], which can easily occur as it is difficult for patients to hold their breath for the full duration of pCT scanning. We therefore used a nonrigid image registration algorithm to preprocess the pCT data. The algorithm was based on the diffeomorphic demons approach, modified by use of normalized gradient fields (NGF) to handle intensity changes caused by contrast uptake. The registration algorithm uses a multiresolution framework with three levels (128 × 128 × 8, 256 × 256 × 8, 512 × 512 × 8), the final spacing being equal to the original voxel spacing. The maximum number of iterations for each level was 25, and the standard deviation of the Gaussian smoothing kernel was 2.8 mm, 1.4 mm, and 0.7 mm at the different resolution levels. Further details have been provided by Papiez et al [27].
The motioncorrected pCT data was then processed voxelbyvoxel using the commercial GE Perfusion 4D software (GE Healthcare, Milwaukee, USA). Voxelbyvoxel tumor blood flow information was obtained by fitting the Adiabatic Approximation to the Tissue Homogeneity (AATH) model [28] to the pCT TACs of each voxel, which describe the variation of voxel Xray attenuation coefficient with time. Similarly to the compartment models used in the PET kinetics analysis, the AATH model describes the timecourse of attenuation, and thus of iodine uptake, as the convolution of an iodine input function, obtained from a VOI drawn in the center of the descending aorta, with a residue function containing fittable parameters. A schematic of the AATH model is shown in Fig. 3: unlike the PET models, blood flowing through the tumor is considered to have a finite transit time, with contrast exchanged between intra and extravascular spaces only at the venous outlet.
Voxelbyvoxel values of blood flow, BF, were taken from the resulting parametric images and averaged over tumor volumes. Then, the averaged BF values for each tumor volume were compared to measures obtained from PET kinetics analysis. When the whole tumor volume exceeded the 4cm axial length of the pCT scans, as shown in Fig. 4, the PET kinetic analysis was repeated just for the tumor subvolume lying within the pCT field of view, allowing results obtained from pCT perfusion and PET kinetics analyses to be meaningfully compared.
The K_{1} rateconstant obtained from PET kinetic modeling is conceptually linked to BF via
in which the extraction fraction E for a cylindrical capillary is given by [29, 30]
where P is capillary permeability, S the surface area per unit volume, and PS the permeability surface area product. For highly permeable tracers such as FMISO [31], PS is much greater than BF and the extraction fraction is close to 1. Consequently, BF measurements obtained from pCT scans should be approximately equal to K_{1} derived from FMISO PET kinetic modeling. We have therefore determined the Pearson r coefficients of correlation between pCT tumor mean BF values and K_{1} values obtained from 2C3K and 3C5K compartment model fits to the dPET data. Correlations were also evaluated between BF values and static FMISO uptakes at each timepoint in the dynamic series of images, and between BF and the average FMISO uptakes over the first 2 min postinjection.
Results
Quality of compartment model fits to FMISO TACs
The numbers of whole tumor FMISO TACs for which fits of each compartment model passed the runstest are listed in Table 2, together with total AIC, BIC, and MSEP scores for the different models summed over all TACs, and the numbers of TACs for which each model achieved the lowest scores. Runstest results are presented individually for each whole tumor TAC in Additional file 1: Table S1 with corresponding AIC, BIC, and MSEP scores detailed in Additional file 2: Table S2.
The three and fourtissue models passed the runstest for 83–100% of whole tumor TACs, whereas twotissue model fits passed for only 0–20% [32]. Summed AIC, BIC, and MSEP scores were much lower for three and fourtissue models than for twotissue models, which lacked the flexibility to describe the data well. Fits of the 2C3K, 2C4K, and 3C5K models to two example TACs (4 and 24) are plotted in Fig. 5.
The irreversible threetissue model, 3C5K, achieved the lowest AIC, BIC, and MSEP scores totaled over all tumor TACs, and the lowest scores for most (21–24) individual TACs. Individual scores were generally a little higher for the 3C6K, 4C7K, and 4C8K models, their additional complexity usually being unnecessary to describe the TAC data.
Table 3 lists estimates of parameter accuracy and precision obtained from the statistical simulations, which used fits of the 3C5K and 3C6K models to measured FMISO TACs to represent the groundtruth, the 3C5K model offering the best description of the TACs according to the AIC, BIC, and MSEP measures. Parameter values obtained from fits of the 2C4K model to simulated noisy TACs had large mean biases and total uncertainties, irrespective of which threetissue groundtruth model was used. Mean biases and uncertainties of fitted 2C3K model parameters were not so large as for 2C4K, but were still considerably larger than those estimated for threecompartment models.
For parameters ν_{B}, K_{1}, k_{2}, k_{3}, and k_{4}, mean biases and total uncertainties were similar for 3C5K and 3C6K model fits, regardless of which threetissue model was used to represent the groundtruth. For the k_{5} and k_{flux} parameters, however, mean biases and total uncertainties were notably lower for 3C5K than for 3C6K model fits when the groundtruth was represented by the 3C5K model. Total uncertainties on 3C5K fits remained less than those on 3C6K fits even when the 3C6K model was used to represent the groundtruth, although in this circumstance 3C5K fit parameters had slightly higher mean biases than 3C6K fits.
Overall, the 3C5K model provided the most precise estimates of FMISO uptake kinetics according to statistical simulations, and the model’s accuracy was only surpassed by 3C6K when this reversible model was also considered to represent the groundtruth, despite the known irreversibility of FMISO binding.
Correlations between K _{1} and BF parameters obtained from FMISO dPET and pCT
Tumor K_{1} values obtained from fits of the 3C5K and 2C3K models to the FMISO dPET data are plotted in Fig. 6 against BF values independently obtained from pCT analysis. The 3C5Kbased K_{1} values were strongly correlated with BF (Pearson r coefficient = 0.81), whereas 2C3Kbased K_{1} values were less strongly correlated (r = 0.68). Pearson r coefficients of correlation between BF and static FMISO tumor uptake in frames collected at different times are plotted in Fig. 7, the maximum correlation (r = 0.70) being obtained at 2min postinjection.
Discussion
Whole tumor FMISO TACs obtained from dPET scans of advanced stage NSCLC patients were described better by an irreversible threetissue compartment model, 3C5K, than by other compartment models we studied, according to information criterion and crossvalidation scores, and statistical simulations. Total information criterion and crossvalidation scores were much worse for simpler twotissue compartment models and slightly worse for the reversible threetissue model, 3C6K, and for fourtissue models whose additional complexity was unnecessary. In statistical simulation studies, total uncertainties calculated for fitted 3C5K model parameter values were consistently lower than those found for twotissue compartment model fits, and a little lower than for 3C6K fits, even when the 3C6K model was used to represent the groundtruth.
For five of the measured whole tumor TACs, the 3C5K model fits did not pass a runstest whereas the fourtissue compartment model fits did. For these particular TACs, we therefore carried out further statistical simulations, using 3C5K and 4C7K models as groundtruth. Even for these specific cases, fits of the 3C5K model provided more precise estimates of groundtruth kinetics values than did 4C7K fits, regardless of the groundtruth model used in the simulations.
A strong correlation (r = 0.81) was found between the K_{1} parameter values of the 3C5K model fits to the FMISO TACs and BF values independently obtained from pCT. The K_{1} values obtained from 2C3K model fits correlated less strongly (albeit not significantly less strongly) with BF (r = 0.68), lending weight to the results indicating that the 3C5K model describes whole tumor FMISO kinetics better than 2C3K. The K_{1} values obtained from 3C5K model fits were also more strongly correlated with BF than were whole tumor FMISO uptake values at times ranging from tracer injection to 4 h later (maximum correlation r = 0.70).
FMISO dPET imaging is used to measure hypoxia, and since hypoxia is related to perfusion, pCT perfusion scans are sometimes collected as well. Since BF is strongly correlated with the K_{1} values obtained from 3C5K model fits to FMISO dPET data, blood flow could potentially be estimated directly from the K_{1} values obtained from the FMISO images, rather than from pCT, thus saving time, money and the pCT radiation dose, and generating BF data over the 15 cm axial width of modern PET scanner fieldsofview, rather than 4cm axial width typical of CT scanners used in cinemode. Blood flow could also be estimated from FDG dPET data if available (not a hypoxia tracer), since BF values obtained from ^{15}Olabeled water dPET studies have previously been shown to strongly correlate with parameter values obtained from model fits to the first 2 min of FDG dPET scans (r = 0.86) [33].
The goldstandard method for determining input functions is direct arterial line sampling. However, we have used imagederived input functions (IDIFs) calculated from mean tracer activity concentrations within volumes drawn in the descending aorta, both for patient comfort and safety, and because good agreement has been demonstrated between directly sampled input functions and IDIFs obtained from the descending aorta [11].
In this study, we have assessed the performance of PET kinetic models in terms of information criteria scores, crossvalidation measures, statistical simulations, and strengths of correlations with an independent measure of perfusion. The imaging protocol used in this work is demanding, as we are currently investigating whether shorter protocols can provide adequate rateconstant estimates. In a recently opened study (Atovaquone as Tumor HypOxia Modifier, NCT02628080), surgically treated NSCLC patients are being imaged using dynamic FMISO PET prior to tumor excision, allowing us to compare FMISO images and parametric maps directly with maps of histopathology obtained from excised tumor slices.
Conclusions
Timecourses of whole tumor FMISO uptake in patients with advanced stage NSCLC were described better by an irreversible threetissue compartment model, 3C5K, than by other two, three, or fourtissue compartment models investigated. Fits of this model also provided the most precise estimates of FMISO uptake kinetics according to simulation studies. Further evidence for the utility of the 3C5K model was provided by the observation of a strong correlation (r = 0.81) between fitted values of its K_{1} parameter and blood flow values obtained independently from perfusion CT imaging, a stronger correlation than that between blood flow and K_{1} values obtained from 2C3K model fits, or between blood flow and tumor FMISO uptake in static scans taken at a range of times from immediately postinjection to 4 h later.
Abbreviations
 2C3K:

Irreversible twotissue compartment model
 3C5K:

Irreversible threetissue compartment model
 AATH:

Adiabatic Approximation to the Tissue Homogeneity
 AIC:

Akaike information criterion
 BF:

Blood flow
 BIC:

Bayesian information criterion
 CT:

Computed tomography
 dPET:

Dynamic PET
 FDG:

^{18}Ffluorodeoxyglucose
 FMISO:

^{18}Ffluoromisonidazole
 IDIF:

Imagederived input function
 MB:

Mean bias
 MSEP:

Mean of the squared error of prediction
 NGF:

Normalized gradient fields
 NSCLC:

Nonsmall cell lung cancer
 pCT:

Perfusion CT
 PET:

Positron emission tomography
 RT:

Radiotherapy
 TACs:

Timeactivitycurves
 VOI:

Volumeofinterest
References
 1.
Rasey JS, Koh WJ, Grierson JR, Grunbaum Z, Krohn KA. Radiolabeled fluoromisonidazole as an imaging agent for tumor hypoxia. Int J Radiat Oncol Biol Phys. 1989;17:985–91.
 2.
Casciari J, Graham M, Rasey J. A modeling approach for quantifying tumor hypoxia with [F18]fluoromisonidazole PET timeactivity data. Med Phys. 1995;22:1127–39.
 3.
Eschmann S, Paulsen F. Prognostic impact of hypoxia imaging with 18Fmisonidazole PET in nonsmall cell lung cancer and head and neck cancer before radiotherapy. J Nucl Med. 2005;46:253–60.
 4.
Koh WJ, Rasey JS, Evans ML, et al. Imaging of hypoxia in human tumors with [F18]fluoromisonidazole. Int J Radiat Oncol Biol Phys. 1992;22:199–212.
 5.
Thorwarth D, Eschmann SM, Scheiderbauer J, Paulsen F, Alber M. Kinetic analysis of dynamic 18Ffluoromisonidazole PET correlates with radiation treatment outcome in headandneck cancer. BMC Cancer. 2005;5:152.
 6.
Axel L. Cerebral blood flow determination by rapidsequence computed tomography: theoretical analysis. Radiology. 1980;137(3):679–86.
 7.
Grüner JM, Paamand R, Højgaard L, Law I. Brain perfusion CT compared with15OH2OPET in healthy subjects. EJNMMI Res. 2011;1(1):28.
 8.
Mullani NA, Herbst RS, O’Neil RG, Gould KL, Barron BJ, Abbruzzese JL. Tumor blood flow measured by PET dynamic imaging of firstpass F18FDG uptake: a comparison with O15labeled watermeasured blood flow. J Nucl Med. 2008;49(4):517–23.
 9.
Fokas E, Im JH, Hill S, et al. Dual inhibition of the PI3K/mTOR pathway increases tumor radiosensitivity by normalizing tumor vasculature. Cancer Res. 2012;72:239–48.
 10.
McGowan DR, Macpherson RE, Hackett SL, et al. ^{18} Ffluoromisonidazole uptake in advanced stage nonsmall cell lung cancer: a voxelbyvoxel PET kinetics study. Med Phys. 2017;44:4665–76. https://doi.org/10.1002/mp.12416
 11.
van der Weerdt AP, Klein LJ, Boellaard R, Visser CA, Visser FC, Lammertsma AA. Imagederived input functions for determination of MRGlu in cardiac (18)FFDG PET scans. J Nucl Med. 2001;42(11):1622–9.
 12.
Muzi M, O’Sullivan F, Mankoff DA, et al. Quantitative assessment of dynamic PET imaging data in cancer imaging. Magn Reson Imaging. 2012;30:1203–15.
 13.
Li F, Joergensen J, Hansen A, Kjaer A. Kinetic modeling in PET imaging of hypoxia. Am J Nucl Med Mol Imaging. 2014;4:490–506.
 14.
Sokoloff L, Reivich M, Kennedy C, et al. The [14C]deoxyglucose method for the measurement of local cerebral glucose utilization: theory, procedure, and normal values in the conscious and anesthetized albino rat. J Neurochem. 1977;28:897–916.
 15.
Bertoldo A, Peltoniemi P, Oikonen V, Knuuti J, Nuutila P, Cobelli C. Kinetic modeling of [(18)F]FDG in skeletal muscle by PET: a fourcompartment fiverateconstant model. Am J Physiol Endocrinol Metab. 2001;281:E524–36.
 16.
Feng D, Huang SC, Wang X. Models for computer simulation studies of input functions for tracer kinetic modeling with positron emission tomography. Int J Biomed Comput. 1993;32:95–110.
 17.
Liu D, Chalkidou A, Landau DB, Marsden PK, Fenwick JD. 18FFLT uptake kinetics in head and neck squamous cell carcinoma: a PET imaging study. Med Phys. 2014;41:41911.
 18.
Press WH, Teukolsky SA, Vetterling WT, Flannery BP. Numerical recipes in C: the art of scientific computing. 2nd ed. New York: Cambridge University Press; 1992. p. 656–706. Ch.15
 19.
Wald A, Wolfowitz J. On a test whether two samples are from the same population. Ann Math Stat. 1940;11:147–62.
 20.
Cobelli C, Foster D, Toffolo G. Tracer kinetics in biomedical research: from data to model. New York: Kluwer Academic; 2001.
 21.
Akaike H. A new look at the statistical model identification. Autom Control IEEE Trans. 1974;19:716–23.
 22.
Burnham KP. Multimodel inference: understanding AIC and BIC in model selection. Sociol Methods Res. 2004;33:261–304.
 23.
Picard RR, Cook RD. Crossvalidation of regression models. J Am Stat Assoc. 1984;79:575–83.
 24.
Wang W, Georgi JC, Nehmeh S, et al. Evaluation of a compartmental model for estimating tumor hypoxia via FMISO dynamic PET imaging. Phys Med Biol. 2009;54:3083–99.
 25.
Guo H, Renaut R, Chen K, Reiman E. FDGPET parametric imaging by total variation minimization. Comput Med Imaging Graph. 2009;33:295–303.
 26.
Muzi M, Mankoff DA, Grierson JR, Wells JM, Vesselle H, Krohn KA. Kinetic modeling of 3′deoxy3′fluorothymidine in somatic tumors: mathematical studies. J Nucl Med. 2005;46:371–80.
 27.
Papież BW, Heinrich MP, Fehrenbach J, Risser L, Schnabel JA. An implicit slidingmotion preserving regularisation via bilateral filtering for deformable image registration. Med Image Anal. 2014;18(8):1299–311.
 28.
St Lawrence KS, Lee TY. An adiabatic approximation to the tissue homogeneity model for water exchange in the brain: II. Experimental validation. J Cereb Blood Flow Metab. 1998;18:1378–85.
 29.
Renkin EM. Transport of potassium42 from blood to tissue in isolated mammalian skeletal muscles. Am J Physiol. 1959;197(6):1205–10.
 30.
Crone C. The permeability of capillaries in various organs as determined by use of the “Indicator diffusion” method. Acta Physiol Scand. 1963;58:292–305.
 31.
Kelly CJ, Brady M. A model to simulate tumour oxygenation and dynamic [18F]Fmiso PET data. Phys Med Biol. 2006;51(22):5859–73.
 32.
Raftery A. Bayesian model selection in social research. Sociol Methodol. 1995;25:111–63.
 33.
Mullani NA, Herbst RS, O’Neil RG, Gould KL, Barron BJ, Abbruzzese JL. Tumor blood flow measured by PET dynamic imaging of firstpass 18FFDG uptake: a comparison with 15Olabeled watermeasured blood flow. J Nucl Med. 2008;49(4):517–23.
Acknowledgements
The authors would like to acknowledge the use of the University of Oxford Advanced Research Computing (ARC) facility in carrying out this work. https://doi.org/10.5281/zenodo.22558
Funding
DM is funded by a National Institute for Health Research (NIHR)/Health Education England (HEE) Clinical Lectureship (ICACL201602009). This paper presents independent research funded by the NIHR. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, HEE, or the Department of Health. GH is supported by a Cancer Research UK Clinician Scientist Awards (C34326/A13092 and C34326/A19590). JF was supported by a Cancer Research UK Career Development Fellowship (C17203).
The BKM120 trial is sponsored by the University of Oxford, managed by OCTRU, and supported by Cancer Research UK (C34326/A15163), the Oxford ECMC, the CRUK EPSRC Oxford Cancer Imaging Centre, the CRUK Oxford Centre, CTRad, and Novartis. The Cancer Research UK/MRC Oxford Institute for Radiation Oncology is supported by core grants from the Medical Research Council and Cancer Research UK.
Availability of data and materials
The analyzed data is available once the final trial results are published subject to reasonable request and approval by the BKM120 clinical trial office.
Author information
Author notes
Affiliations
Contributions
DM drafted the manuscript and performed data analysis. DM, MS, RM, and GH contributed to data acquisition. All authors contributed to data processing and manuscript revision. All authors read and approved the final manuscript.
Corresponding author
Correspondence to Daniel R. McGowan.
Ethics declarations
Ethics approval and consent to participate
The BKM120 clinical trial (NCT02128724) has been approved by the local ethics committee (Oxford B, reference 12/SC/0674), and signed informed consent obtained from all patients.
Consent for publication
Not applicable
Competing interests
The authors declare that they have no competing interests.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Additional files
Additional file 1:
Table S1. TACbyTAC WaldWolfowitz runstest results for fits of the different models (ticks indicate runstest passes) (DOCX 38 kb)
Additional file 2:
Table S2. Individual AIC, BIC and MSEP scores for fits of the various models to each TAC. The lowest AIC, BIC, and MSEP scores have been underlined for each TAC, indicating the best model according to that measure (DOCX 46 kb)
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Received
Accepted
Published
DOI
Keywords
 FMISO
 NSCLC
 Dynamic PET
 Kinetics analysis
 Perfusion CT