An introduction to PET and SPECT neuroreceptor quantification models.

PET and SPECT using appropriate radioligands allow imaging of certain critical components of neurotransmission such as presynaptic transporters and postsynaptic receptors in living human brains. PET and SPECT data are commonly analyzed by applying tracer kinetic models. These modeling approaches assume a compartmental system and derive the outcome measure called the binding potential, which reflects the densities of transporters or receptors in a brain region of interest. New models are often noninvasive in that they do not require arterial blood sampling. In this review, the concept and principles of tracer kinetic modeling are introduced and commonly used PET and SPECT neuroreceptor quantification models are discussed.     

Transvascular and interstitial transport in rat hepatocellular carcinomas: dynamic contrast-enhanced MRI assessment with low- and high-molecular weight agents

PURPOSE: To assess which MRI-derived kinetic parameters reflect decreased transvascular and interstitial transport when low- and high-molecular-weight agents are used in rat hepatocellular carcinomas. MATERIALS AND METHODS: Dynamic MRI after injection of a low-molecular-weight contrast agent of 0.56 kDa (Gd-DOTA, gadoterate) and two high-molecular-weight contrast agents of 6.47 kDa (P792, gadomelitol) and 52 kDa (P717, carboxymethyldextran Gd-DOTA) was performed in rats with chemically induced hepatocellular carcinomas. The data were analyzed with the Kety compartmental model, the extended Kety compartmental model in which it is assumed that the tissue voxels contain a vascular component, and the St Lawrence and Lee distributed-parameter model. RESULTS: The extravascular extracellular space accessible to the contrast agent v(e) and the extraction fraction E decreased with increasing molecular weight of the contrast agent. In contrast, the volume transfer constant Ktrans did not differ significantly when low- or high-molecular-weight agents were used. CONCLUSION: In this animal model the results suggest that the accessible extravascular extracellular space and the extraction fraction are more sensitive indicators of decreased transvascular and interstitial transport with high-molecular-weight agents than the volume transfer constant, which is a lumped representation of blood flow and permeability.     

Arterial input function measurement without blood sampling using a beta-microprobe in rats.

The evaluation of every new radiotracer involves pharmacokinetic studies on small animals to determine its biodistribution and local kinetics. To extract relevant biochemical information, time-activity curves for the regions of interest are mathematically modeled on the basis of compartmental models that require knowledge of the time course of the tracer concentration in plasma. Such a time-activity curve, usually termed input function, is determined in small animals by repeated blood sampling and subsequent counting in a well counter. The aim of the present work was to propose an alternative to blood sampling in small animals, since this procedure is labor intensive, exposes the staff to radiation, and leads to an important loss of blood, which affects hematologic parameters. METHODS: Monte Carlo simulations were performed to evaluate the feasibility of measuring the arterial input function using a positron-sensitive microprobe placed in the femoral artery of a rat. The simulation results showed that a second probe inserted above the artery was necessary to allow proper subtraction of the background signal arising from tracer accumulation in surrounding tissues. This approach was then validated in vivo in 5 anesthetized rats. In a second set of experiments, on 3 rats, a third probe was used to simultaneously determine 18F-FDG accumulation in the striatum. RESULTS: The high temporal resolution of the technique allowed accurate determination of the input function peak after bolus injection of 18F-FDG. Quantitative input functions were obtained after normalization of the arterial time-activity curve for a late blood sample. In the second set of experiments, compartmental modeling was achieved using either the blood samples or the microprobe data as the input function, and similar kinetic constants were found in both cases. CONCLUSION: Although direct quantification proved difficult, the microprobe allowed accurate measurement of arterial input function with a high temporal resolution and no blood loss. The technique, because offering adequate sensitivity and temporal resolution for kinetic measurements of radiotracers in the blood compartment, should facilitate quantitative modeling for radiotracer studies in small animals.     

PET kinetic analysis: error consideration of quantitative analysis in dynamic studies

Positron emission tomography dynamic studies have been performed to quantify several biomedical functions. In a quantitative analysis of these studies, kinetic parameters were estimated by mathematical methods, such as a nonlinear least-squares algorithm with compartmental model and graphical analysis. In this estimation, the uncertainty in the estimated kinetic parameters depends on the signal-to-noise ratio and quantitative analysis method. This review describes the reliability of parameter estimates for various analysis methods in reversible and irreversible models.     

Fundamentals of tracer kinetics for dynamic contrast-enhanced MRI

Tracer kinetic methods employed for quantitative analysis of dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) share common roots with earlier tracer studies involving arterial-venous sampling and other dynamic imaging modalities. This article reviews the essential foundation concepts and principles in tracer kinetics that are relevant to DCE MRI, including the notions of impulse response and convolution, which are central to the analysis of DCE MRI data. We further examine the formulation and solutions of various compartmental models frequently used in the literature. Topics of recent interest in the processing of DCE MRI data, such as the account of water exchange and the use of reference tissue methods to obviate the measurement of an arterial input, are also discussed. Although the primary focus of this review is on the tracer models and methods for T(1) -weighted DCE MRI, some of these concepts and methods are also applicable for analysis of dynamic susceptibility contrast-enhanced MRI data.     

Kinetic modeling in positron emission tomography.

Most PET kinetic modeling approaches have at their basis a compartmental model that has first-order, constant coefficients. The present article outlines the one-, two-, and three-compartment models used to measure cerebral blood flow, cerebral glucose metabolism, and receptor binding, respectively. The number of compartments of each model is based on specific knowledge of the physiological and/or biochemical compartments into which the tracer distributes. Additional physical and biochemical properties of the tracer distribution are considered in specifying the use of first-order rate constants. For example, in cerebral blood flow and receptor binding studies transport across the blood-brain barrier by diffusion can be modeled as a first-order process. A saturable carrier-mediated process or saturable enzyme catalyzed reaction, when tracer doses of the labeled substrate are used and the natural substrate is in steady-state, also results in first-order rate constants, as in glucose metabolism studies. The rate of ligand binding, on the other hand, depends on the concentrations of both substrate and available receptors. In order to appropriately model the reaction as pseudo first-order during a specified experimental interval, protocols are carefully designed to assure that the number of available binding sites remains approximately constant throughout the given interval. A broad array of scanning protocols is employed for kinetic analyses. These include single-scan approaches, which function like their autoradiographic counterparts in animal studies and are often called "autoradiographic" methods, which allow estimation of a single parameter. Dynamic scanning to obtain the time course of tissue activity allows simultaneous estimation of multiple parameters. Scanning may be conducted during a period of tracer uptake or after attainment of steady-state conditions. All quantitative modeling approaches share the common requirement that an arterial input function be measured or an appropriate surrogate be found. A vast array of methods is available for estimation of model parameters, both micro and macro. In the final analysis, it is the interaction among all elements of the PET study, including careful tracer selection, model specification, experimental protocol design, and sound parameter estimation methods, that determines the quantitative accuracy of the estimates of the physiological or biochemical process under study.     

Dynamic small-animal PET imaging of tumor proliferation with 3'-deoxy-3'-18F-fluorothymidine in a genetically engineered mouse model of high-grade gliomas.

3'-Deoxy-3'-(18)F-fluorothymidine ((18)F-FLT), a partially metabolized thymidine analog, has been used in preclinical and clinical settings for the diagnostic evaluation and therapeutic monitoring of tumor proliferation status. We investigated the use of (18)F-FLT for detecting and characterizing genetically engineered mouse (GEM) high-grade gliomas and evaluating the pharmacokinetics in GEM gliomas and normal brain tissue. Our goal was to develop a robust and reproducible method of kinetic analysis for the quantitative evaluation of tumor proliferation. METHODS: Dynamic (18)F-FLT PET imaging was performed for 60 min in glioma-bearing mice (n = 10) and in non-tumor-bearing control mice (n = 4) by use of a dedicated small-animal PET scanner. A 3-compartment, 4-parameter model was used to characterize (18)F-FLT kinetics in vivo. For compartmental analysis, the arterial input was measured by placing a region of interest over the left ventricular blood pool and was corrected for partial-volume averaging. The (18)F-FLT "trapping" and tissue flux model parameters were correlated with measured uptake (percentage injected dose per gram [%ID/g]) values at 60 min. RESULTS: (18)F-FLT uptake values (%ID/g) at 1 h in brain tumors were significantly greater than those in control brains (mean +/- SD: 4.33 +/- 0.58 and 0.86 +/- 0.22, respectively; P < 0.0004). Kinetic analyses of the measured time-activity curves yielded independent, robust estimates of tracer transport and metabolism, with compartmental model-derived time-activity data closely fitting the measured data. Except for tracer transport, statistically significant differences were found between the applicable model parameters for tumors and normal brains. The tracer retention rate constant strongly correlated with measured (18)F-FLT uptake values (r = 0.85, P < 0.0025), whereas a more moderate correlation was found between net (18)F-FLT flux and (18)F-FLT uptake values (r = 0.61, P < 0.02). CONCLUSION: A clinically relevant mouse glioma model was characterized by both static and dynamic small-animal PET imaging of (18)F-FLT uptake. Time-activity curves were kinetically modeled to distinguish early transport from a subsequent tracer retention phase. Estimated (18)F-FLT rate constants correlated positively with %ID/g measurements. Dynamic evaluation of (18)F-FLT uptake offers a promising approach for noninvasively assessing cellular proliferation in vivo and for quantitatively monitoring new antiproliferation therapies.     

Comparison of model‐based arterial input functions for dynamic contrast‐enhanced MRI in tumor bearing rats

When using tracer kinetic modeling to analyze dynamic contrast‐enhanced MRI (DCE‐MRI) it is necessary to identify an appropriate arterial input function (AIF). The measured AIF is often poorly sampled in both clinical and preclinical MR systems due to the initial rapid increase in contrast agent concentration and the subsequent large‐scale signal change that occurs in the arteries. However, little work has been carried out to quantify the sensitivity of tracer kinetic modeling parameters to the form of AIF. Using a preclinical experimental data set, we sought to measure the effect of varying model forms of AIF on the extended Kety compartmental model parameters (K^trans, v_e, and v_p) through comparison with the results of experimentally acquired high temporal resolution AIFs. The AIF models examined have the potential to be parameterized on lower temporal resolution data to predict the form of the true, higher temporal resolution AIF. The models were also evaluated through application to the population average AIF. It was concluded that, in the instance of low temporal resolution or noisy data, it may be preferable to use a bi‐exponential model applied to the raw data AIF, or when individual measurements are not available a bi‐exponential model of the average AIF. Magn Reson Med, 2009. © 2009 Wiley‐Liss, Inc.     

Gadolinium-enhanced echo-planar T2-weighted MRI of tumors in the extracranial head and neck: feasibility study and preliminary results using a distributed-parameter tracer kinetic analysis

PURPOSE: To examine the feasibility of first-pass dynamic contrast-enhanced (DCE) T2-weighted MRI of tumors in the extracranial head and neck by applying a distributed-parameter (DP) tracer kinetic model to quantify the perfusion parameters. MATERIALS AND METHODS: A total of 16 patients with primary malignant and benign tumors in the head and neck underwent DCE-MR studies. A spin-echo (SE) echo-planar-imaging (EPI) MR-sequence was applied for first-pass DCE-T2-weighted imaging. The data were postprocessed applying a DP tracer kinetic model that accounts for capillary-tissue exchange. Region-of-interest (ROI) analysis was performed in the tumor sites and the adjacent normal tissue. Blood flow (F), intravascular blood volume (v(1)), extravascular extracellular volume (v(2)), difference in bolus arrival time between arterial input and tissue (t(0)), intravascular mean transit time (t(1)), permeability (PS), and extraction ratio (E) maps were generated for each patient. RESULTS: All perfusion values in the tumor sites were significantly different (0.000 < or = P < or = 0.01) than those in the normal muscle tissue. The median perfusion values in the tumor tissue were: F = 150.5 mL/minute/100 g, v(1) = 11.0 mL/100 g, v(2) = 31.5 mL/100 g, t(0) = 4.5 seconds, t(1) = 8.0 seconds, PS = 96.0 mL/minute/100 g, and E = 32.5. CONCLUSION: EPI-T2-weighted DCE-MR in head and neck tumors as well as quantification of the perfusion values using DP model physiologic imaging was feasible and the promising initial results have encourages further validation studies in the future.     

Comparison of the performance of tracer kinetic model-driven registration for dynamic contrast enhanced MRI using different models of contrast enhancement

RATIONALE AND OBJECTIVES: The quantitative analysis of dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) data is subject to model fitting errors caused by motion during the time-series data acquisition. However, the time-varying features that occur as a result of contrast enhancement can confound motion correction techniques based on conventional registration similarity measures. We have therefore developed a heuristic, locally controlled tracer kinetic model-driven registration procedure, in which the model accounts for contrast enhancement, and applied it to the registration of abdominal DCE-MRI data at high temporal resolution. MATERIALS AND METHODS: Using severely motion-corrupted data sets that had been excluded from analysis in a clinical trial of an antiangiogenic agent, we compared the results obtained when using different models to drive the tracer kinetic model-driven registration with those obtained when using a conventional registration against the time series mean image volume. RESULTS: Using tracer kinetic model-driven registration, it was possible to improve model fitting by reducing the sum of squared errors but the improvement was only realized when using a model that adequately described the features of the time series data. The registration against the time series mean significantly distorted the time series data, as did tracer kinetic model-driven registration using a simpler model of contrast enhancement. CONCLUSION: When an appropriate model is used, tracer kinetic model-driven registration influences motion-corrupted model fit parameter estimates and provides significant improvements in localization in three-dimensional parameter maps. This has positive implications for the use of quantitative DCE-MRI for example in clinical trials of antiangiogenic or antivascular agents.     

Liver kinetics of glucose analogs measured in pigs by PET: importance of dual-input blood sampling.

Metabolic processes studied by PET are quantified traditionally using compartmental models, which relate the time course of the tracer concentration in tissue to that in arterial blood. For liver studies, the use of arterial input may, however, cause systematic errors to the estimated kinetic parameters, because of ignorance of the dual blood supply from the hepatic artery and the portal vein to the liver. METHODS: Six pigs underwent PET after [15O]carbon monoxide inhalation, 3-O-[11C]methylglucose (MG) injection, and [18F]FDG injection. For the glucose scans, PET data were acquired for 90 min. Hepatic arterial and portal venous blood samples and flows were measured during the scan. The dual-input function was calculated as the flow-weighted input. RESULTS: For both MG and FDG, the compartmental analysis using arterial input led to systematic underestimation of the rate constants for rapid blood-tissue exchange. Furthermore, the arterial input led to absurdly low estimates for the extracellular volume compared with the independently measured hepatic blood volume of 0.25 +/- 0.01 mL/mL (milliliter blood per milliliter liver tissue). In contrast, the use of a dual-input function provided parameter estimates that were in agreement with liver physiology. Using the dual-input function, the clearances into the liver cells (K1 = 1.11 +/- 0.11 mL/min/mL for MG; K1 = 1.07 +/- 0.19 mL/min/mL for FDG) were comparable with the liver blood flow (F = 1.02 +/- 0.05 mL/min/mL). As required physiologically, the extracellular volumes estimated using the dual-input function were larger than the hepatic blood volume. The linear Gjedde-Patlak analysis produced parameter estimates that were unaffected by the choice of input function, because this analysis was confined to time scales for which the arterial-input and dual-input functions were very similar. CONCLUSION: Compartmental analysis of MG and FDG kinetics using dynamic PET data requires measurements of dual-input activity concentrations. Using the dual-input function, physiologically reasonable parameter estimates of K1, k2, and Vp were obtained, whereas the use of conventional arterial sampling underestimated these parameters compared with independent measurements of hepatic flow and hepatic blood volume. In contrast, the linear Gjedde-Patlak analysis, being less informative but more robust, gave similar parameter estimates (K, V) with both input functions.     

Evaluation of treatment response of cilengitide in an experimental model of breast cancer bone metastasis using dynamic PET with 18F-FDG

The purpose of this study was the assessment of the feasibility of dynamic positron emission tomography (PET) studies with fluorine-18 fluorodeoxyglucose ((18)F-FDG) to quantify effects of the cyclic Arg-Gly-Asp peptide cilengitide, which targets the ανβ 3 and ανβ 5 integrin receptors in rats with breast cancer bone metastases. Rats were inoculated with MDA-MB-231 breast cancer cells, followed by the development of lytic lesions in the hind leg. Rats with lytic lesions were treated with cilengitide five times weekly on a continuous basis from days 30 to 55 after tumor cell inoculation. Dynamic PET studies with (18)F-FDG were performed in untreated (n=9), controlled (n=4) and treated rats (n=6). The data were assessed using learning-machine two-tissue compartmental analysis. The (18)F-FDG kinetic parameters obtained by two-tissue compartmental model learning-machine showed significant differences when individual parameters were compared between the control group and treated animals. Quantitative assessment of the tracer kinetics and the application of classification analysis to the data provided us with evidence to identify those tumors that demonstrated effect of cilengitide treatment. The transport rate K1 and the phosphorylation rate k3 were significantly different (P=0.033 and 0.038, respectively). Classification analysis based on support vector machines ranking feature elimination of the combination of PET parameters revealed an overall accuracy of 80.0% between treated animals and the control group. We were able to identify 83.3% treated animals compared with the control group based on k2 and VB. In conclusion, the results revealed that cilengitide treatment of experimental breast cancer bone metastases had a significant therapeutic impact on (18)F-FDG kinetics.     

Temporal behavior of peripheral organ distribution volume in mammillary systems. I. A new tool for in vivo tracer kinetic analysis

In complex mammillary compartmental systems, the kinetic solutions for central and peripheral compartments are sums of too many exponentials to be accurately analyzed without very sophisticated mathematical tools. Our data show the peripheral organ distribution volume (PODV) kinetics to exhibit systematic time behavior depending on its mode of relation with plasma: linear increase for irreversible transfer, uniexponential function growing toward as asymptotic value for reversible transfer. Statistical analysis of our kinetic data shows that no other significant information can be extracted at least inside the time and statistical noise limits of our investigation. After intravenous injection of a diffusible tracer, the total activity in any region of interest (ROI) in the body is the sum of various components and, under certain conditions, PODV transformation easily allows their separation. Our simple non-compartmental model provides a useful tool for quantitative tracer analysis in nuclear medicine.     

Free-breathing motion compensation using template matching: a technique allowing for tracer kinetic modeling of liver metastases

Modeling tracer kinetics from dynamic magnetic resonance imaging (dMRI) to understand microvascular characteristics typically requires acquisitions longer than 1 breath-hold. This has limited the application of dMRI in assessment of the upper abdomen. Here we present a template-based motion correction strategy for dMRI of liver metastases based on the correlation coefficient (CC), originally developed for tracking coronary arteries. This postprocessing method allows patient free breathing during sagittal dMRI acquisition and allows a more precise parametric mapping using tracer kinetic models. In a study of 6 subjects, a 64 x 64 template was accurately tracked retrospectively with mean CC = 0.72 +/- 0.07. Mean superior-inferior displacement tracked was 1.82 +/- 1.20 pixels, whereas mean anterior-posterior displacement was 7.72 +/- 4.58 pixels. Application of the CC method significantly improved the global fit (chi2) of a tracer kinetic model throughout tumor regions. Therefore, use of the CC postprocessing method for dMRI scans can improve the precision of dMRI tracer kinetic models.     

Comparison of MRI and positron emission tomography for measuring myocardial perfusion reserve in healthy humans.

Myocardial perfusion reserve (MPR, defined as the ratio of the maximum myocardial blood flow (MBF) to the baseline) is an indicator of coronary artery disease and myocardial microvascular abnormalities. First-pass contrast-enhanced magnetic resonance imaging (CE-MRI) using gadolinium (Gd)-DTPA as a contrast agent (CA) has been used to assess MPR. Tracer kinetic models based on compartmental analysis of the CA uptake have been developed to provide quantitative measures of MBF by MRI. To study the accuracy of Gd-DTPA first-pass MRI and kinetic modeling for quantitative analysis of myocardial perfusion and MPR during dipyridamole infusion, we conducted a comparison with positron emission tomography (PET) in 18 healthy males (age = 40 +/- 14 years). Five planes were acquired at every second heartbeat with a 1.5T scanner using a saturation recovery turboFLASH sequence. A perfusion-related parameter, the unidirectional influx constant (Ki), was computed in three coronary artery territories. There was a significant correlation for both dipyridamole-induced flow (0.70, P = 0.001) and MPR (0.48, P = 0.04) between MRI and PET. However, we noticed that MRI provided lower MPR values compared to PET (2.5 +/- 1.0 vs. 4.3 +/- 1.8). We conclude that MRI supplemented with tracer kinetic modeling can be used to quantify myocardial perfusion.     

Limited view PET reconstruction of tissue radioactivity maps

This paper proposes a state space method for limited view PET reconstruction. Due to the high-level of noise and data-incompletion, prior knowledge is required to guide PET recovery. The compartmental model is used as an evolution equation to regularize the dynamic reconstruction. The continuous–discrete Kalman filter is adopted to calculate the radioactivity value recursively. With tracer kinetic information as prior, the state space approach can obtain a better result compared with the MLEM algorithm. The identifiability of this method is proved by computer synthetic simulation and real phantom experiment on the Hamamatsu SHR-22000PET scanner.     

Estimating the arterial input function using two reference tissues in dynamic contrast-enhanced MRI studies: fundamental concepts and simulations.

In dynamic contrast-enhanced MRI (DCE-MRI) studies, an accurate knowledge of the arterial contrast agent concentration as a function of time is crucial for the estimation of kinetic parameters. In this work, a novel method for estimating the arterial input function (AIF) based on the contrast agent concentration-vs.-time curves in two different reference tissues is described. It is assumed that the AIFs of the two tissues have the same shape, and that simple models with two or more compartments, and unknown kinetic parameters, can describe their tracer concentration-vs.-time curves. Based on the principle of self-consistency, one can relate the two tracer concentration-vs.-time curves to estimate their common underlining AIF, together with the kinetic parameters of the two tissues. In practice, the measured concentration-vs.-time curves have noise, and the AIFs of the two tissues are not exactly the same due to different dispersion effects. These factors will produce errors in the AIF estimate. Simulation studies show that despite the two error sources, the double-reference-tissue method provides reliable estimates of the AIF.     

Effect of arteriovenous difference in measuring renal function by single-injection plasma clearance: role of bolus

Error estimates for arteriovenous difference were calculated by two models, a lag time model and a compartmental model, using Tc99m-diethylenetriaminepentaacetic acid (DTPA) plasma clearance curves from 40 subjects and Tc99m-MAG3 (mercaptoacetyltriglycine) curves from 18 subjects. It was found that correcting for the effect of the initial bolus largely cancelled the conventional arteriovenous difference, so that the net error was negligible.     

Graphical analysis of PET data applied to reversible and irreversible tracers

The differential equations of compartmental analysis form the basis of the models describing the uptake of tracers used in imaging studies. Graphical analyses convert the model equations into linear plots, the slopes of which represent measures of tracer binding. The graphical methods are not dependent upon a particular model structure but the slopes can be related to combinations of the model parameters if a model structure is assumed. The input required is uptake data from a region of interest vs time and an input function that can either be plasma measurements or uptake data from a suitable reference region. Graphical methods can be applied to both reversible and irreversibly binding tracers. They provide considerable ease of computation compared to the optimization of individual model parameters in the solution of the differential equations generally used to describe the binding of tracers. Conditions under which the graphical techniques are applicable and some problems encountered in separating tracer delivery and binding are considered. Also the effect of noise can introduce a bias in the distribution volume which is the slope of the graphical analysis of reversible tracers. Smoothing techniques may minimize this problem and retain the model independence. In any case graphical techniques can provide insight into the binding kinetics of tracers in a visual way.     

Early effects of FOLFOX treatment of colorectal tumour in an animal model: assessment of changes in gene expression and FDG kinetics

Purpose  The very early chemotherapeutic effects of the FOLFOX (fluorouracil, folinic acid, oxaliplatin) protocol were assessed in mice implanted with a human colorectal cell line. The aim of this study was to identify changes in gene expression patterns and to detect combinations of PET parameters that may be helpful in identifying treated tumours early after chemotherapy using dynamic PET studies. Methods  A human colorectal cell line (HCT 116) was used in nude mice. Dynamic PET studies were performed in untreated (n = 13) and treated (n = 12) animals. The data were assessed using compartmental and noncompartmental analysis. The removed tumour specimens were assessed by gene array analysis to obtain quantitative information on gene expression. Results  One chemotherapeutic treatment using the FOLFOX protocol resulted in an upregulation of 2,078 gene probes by more than 25%, while 2,254 probes were downregulated following treatment. The gene array data demonstrated primarily an enhancement of genes related to apoptosis. In particular, the apoptosis antigen 1 (APO-1), p21 and the G protein-coupled receptor 87 (G-87) were 2.6- to 3.3-fold upregulated as compared to the expression in untreated animals. There was a 100% separation of untreated and treated animals on the basis of these three genes. The SUV and the FDG kinetic parameters obtained by compartmental and noncompartmental fitting were not significantly different when individual parameters were compared between groups. However, classification analysis of the combination of the PET parameters VB, K1, k3, and influx revealed an overall accuracy of 84%. We were able to identify 91.7% (11/12) of the treated animals and 76.9% (10/13) of the untreated animals correctly using the classification analysis of PET data. Conclusion  Even one chemotherapeutic treatment using FOLFOX has an impact on gene expression and significantly modulates FDG kinetics. Quantitative assessment of the tracer kinetics and the application of classification analysis to the data are promising tools to identify those tumours that demonstrate a chemotherapeutic effect very early following treatment.