Patlak Slope Research Articles

The 2D Hotelling filter - a quantitative noise-reducing principal-component filter for dynamic PET data, with applications in patient dose reduction

BackgroundIn this paper we apply the principal-component analysis filter (Hotelling filter) to reduce noise from dynamic positron-emission tomography (PET) patient data, for a number of different radio-tracer molecules. We furthermore show how preprocessing images with this filter improves parametric images created from such dynamic sequence.We use zero-mean unit variance normalization, prior to performing a Hotelling filter on the slices of a dynamic time-series. The Scree-plot technique was used to determine which principal components to be rejected in the filter process. This filter was applied to [11C]-acetate on heart and head-neck tumors, [18F]-FDG on liver tumors and brain, and [11C]-Raclopride on brain. Simulations of blood and tissue regions with noise properties matched to real PET data, was used to analyze how quantitation and resolution is affected by the Hotelling filter. Summing varying parts of a 90-frame [18F]-FDG brain scan, we created 9-frame dynamic scans with image statistics comparable to 20 MBq, 60 MBq and 200 MBq injected activity. Hotelling filter performed on slices (2D) and on volumes (3D) were compared.ResultsThe 2D Hotelling filter reduces noise in the tissue uptake drastically, so that it becomes simple to manually pick out regions-of-interest from noisy data. 2D Hotelling filter introduces less bias than 3D Hotelling filter in focal Raclopride uptake. Simulations show that the Hotelling filter is sensitive to typical blood peak in PET prior to tissue uptake have commenced, introducing a negative bias in early tissue uptake. Quantitation on real dynamic data is reliable. Two examples clearly show that pre-filtering the dynamic sequence with the Hotelling filter prior to Patlak-slope calculations gives clearly improved parametric image quality. We also show that a dramatic dose reduction can be achieved for Patlak slope images without changing image quality or quantitation.ConclusionsThe 2D Hotelling-filtering of dynamic PET data is a computer-efficient method that gives visually improved differentiation of different tissues, which we have observed improve manual or automated region-of-interest delineation of dynamic data. Parametric Patlak images on Hotelling-filtered data display improved clarity, compared to non-filtered Patlak slope images without measurable loss of quantitation, and allow a dramatic decrease in patient injected dose.

Distribution of Monoamine Oxidase Proteins in Human Brain: Implications for Brain Imaging Studies

Positron emission tomography (PET) imaging of monoamine oxidases (MAO-A: [(11)C]harmine, [(11)C]clorgyline, and [(11)C]befloxatone; MAO-B: [(11)C]deprenyl-D2) has been actively pursued given clinical importance of MAOs in human neuropsychiatric disorders. However, it is unknown how well PET outcome measures for the different radiotracers are quantitatively related to actual MAO protein levels. We measured regional distribution (n=38) and developmental/aging changes (21 hours to 99 years) of both MAOs by quantitative immunoblotting in autopsied normal human brain. MAO-A was more abundant than MAO-B in infants, which was reversed as MAO-B levels increased faster before 1 year and, unlike MAO-A, kept increasing steadily to senescence. In adults, regional protein levels of both MAOs were positively and proportionally correlated with literature postmortem data of MAO activities and binding densities. With the exception of [(11)C]befloxatone (binding potential (BP), r=0.61, P=0.15), correlations between regional PET outcome measures of binding in the literature and MAO protein levels were good (P<0.01) for [(11)C]harmine (distribution volume, r=0.86), [(11)C]clorgyline (λk3, r=0.82), and [(11)C]deprenyl-D2 (λk3 or modified Patlak slope, r=0.78 to 0.87), supporting validity of the latter imaging measures. However, compared with in vitro data, the latter PET measures underestimated regional contrast by ∼2-fold. Further studies are needed to address cause of the in vivo vs. in vitro nonproportionality.

Comparing analysis methods in assessing dynamic dual bolus cardiac magnetic resonance perfusion flow

Background We compare eight reported methods (1-8) for the analysis of cardiac perfusion flow in 3 Tesla MRI on a porcine model. Therefore, an anaesthetized healthy minipig was repeatedly scanned (x5) with a 14 day interval with Turbo FLASH at 3T. The data obtained from the images consists of the temporal course of the arterial input function (AIF) and of the tissue response function (TRF). We compared 8 analysis methods by statistical evaluation of determined perfusion flow. The analysis methods investigated were Fermi function (1), Modelfree Deconvolution (2), Modified Tofts (3), Exchange (4), Uptake (5), Tofts (6), Patlak (7), and Maximum Slope (8).

Quantification of Changes in Skeletal Muscle Amino Acid Kinetics in Adult Humans in Response to Exercise via Positron-emission Tomography with L-[methyl-11C] methionine

Dynamic positron emission tomography (PET) imaging of with L-[methyl- 11 C]methionine (11C-MET) was developed in the late 1990’s to non-invasively estimate skeletal muscle protein synthesis, but no studies have shown that the measurements respond to resistance exercise, which stimulates protein synthesis in humans. Ten healthy women aged 25-75 years underwent a 14-hour fast, followed by unilateral knee extension and flexion exercise and consumption of an 8-ounce serving of fruit juice. Five subjects underwent dynamic 11C-MET PET imaging of the mid-thigh 2-3 hours after exercise and five were imaged 1 hour after exercise. Images were processed to obtain the Patlak slope K i , which describes the fractional extraction rate of 11C-MET into skeletal muscle protein. Additionally, the images were processed with a three-compartment kinetic model to determine rate constants for 11C-MET transport between muscle tissue, protein and plasma. All subjects showed excellent mid-thigh uptake of 11C-MET. Subjects imaged 2-3 hours after exercise showed no unilateral enhancement. However, subjects imaged one hour post-exercise showed an enhancement of 11C-MET uptake in the exercised leg compared to the control leg, corresponding to K i elevations between 3.8% - 31.1%. From the three-compartment analysis, the increased uptake corresponded primarily to an increased rate constant for extraction of 11C-MET from plasma to skeletal muscle tissue. Finally, older subjects tended to have smaller values of K i than the younger subjects. In summary, 11C-MET kinetics is responsive to a unilateral exercise stimulus, and this technique may prove useful to study skeletal muscle amino acid kinetics in response to exercise, aging and other conditions

Image-derived input function in dynamic human PET/CT: methodology and validation with 11C-acetate and 18F-fluorothioheptadecanoic acid in muscle and 18F-fluorodeoxyglucose in brain

PurposeDespite current advances in PET/CT systems, blood sampling still remains the standard method to obtain the radiotracer input function for tracer kinetic modelling. The purpose of this study was to validate the use of image-derived input functions (IDIF) of the carotid and femoral arteries to measure the arterial input function (AIF) in PET imaging. The data were obtained from two different research studies, one using 18F-FDG for brain imaging and the other using 11C-acetate and 18F-fluoro-6-thioheptadecanoic acid (18F-FTHA) in femoral muscles.MethodsThe method was validated with two phantom systems. First, a static phantom consisting of syringes of different diameters containing radioactivity was used to determine the recovery coefficient (RC) and spill-in factors. Second, a dynamic phantom built to model bolus injection and clearance of tracers was used to establish the correlation between blood sampling, AIF and IDIF. The RC was then applied to the femoral artery data from PET imaging studies with 11C-acetate and 18F-FTHA and to carotid artery data from brain imaging with 18F-FDG. These IDIF data were then compared to actual AIFs from patients.ResultsWith 11C-acetate, the perfusion index in the femoral muscle was 0.34±0.18 min−1 when estimated from the actual time–activity blood curve, 0.29±0.15 min−1 when estimated from the corrected IDIF, and 0.66±0.41 min−1 when the IDIF data were not corrected for RC. A one-way repeated measures (ANOVA) and Tukey’s test showed a statistically significant difference for the IDIF not corrected for RC (p<0.0001). With 18F-FTHA there was a strong correlation between Patlak slopes, the plasma to tissue transfer rate calculated using the true plasma radioactivity content and the corrected IDIF for the femoral muscles (vastus lateralis r=0.86, p=0.027; biceps femoris r=0.90, p=0.017). On the other hand, there was no correlation between the values derived using the AIF and those derived using the uncorrected IDIF. Finally, in the brain imaging study with 18F-FDG, the cerebral metabolic rate of glucose (CMRglc) measured using the uncorrected IDIF was consistently overestimated. The CMRglc obtained using blood sampling was 13.1±3.9 mg/100 g per minute and 14.0±5.7 mg/100 g per minute using the corrected IDIF (r 2=0.90).ConclusionCorrectly obtained, carotid and femoral artery IDIFs can be used as a substitute for AIFs to perform tracer kinetic modelling in skeletal femoral muscles and brain analyses.

In vivo measurement of cell proliferation in canine brain tumor using C-11-labeled FMAU and PET

Noncatabolized thymidine analogs are being developed for use in imaging DNA synthesis. We sought to relate a labeling index measured by immunohistochemical staining bromodeoxyuridine (BUdR) technique to the uptake of (11)C 2'-fluoro-5-methyl-1-beta-d-arabinofuranosyluracil (FMAU) measured with positron emission tomography (PET) in a brain tumor model. Adult beagles (n=8) with implanted brain tumors received [(11)C]FMAU and dynamic imaging with arterial sampling. Six dogs were then infused with BUdR (200 mg/m(2)) and sacrificed. Tumor time-activity curves (TACs) obtained from computed-tomography-defined regions of interest were corrected for partial volume effects and crosstalk from brain tissue. Tissue was analyzed for the percentage of tumor volume occupied by viable cells and by viable cells in S-phase as identified by BUdR staining. PET/[(11)C]FMAU and BUdR were compared by linear regression analysis and analysis of variance, as well as by a nonparametric rank correlation test. Tumor standardized uptake values (SUVs) and tumor-to-contralateral-brain uptake ratios at 50 min were 1.6+/-0.4 and 5.5+/-1.2 (n=8; mean+/-S.E.M.), respectively. No (11)C-labeled metabolites were observed in the blood through 60 min. Tumor TACs were well described with a three-compartment/four-parameter model (k(4)=0) and by Patlak analysis. Parametric statistical analysis showed that FMAU clearance from plasma into tumor Compartment 3 (K(FMAU)) was significantly correlated with S-phase percent volume (P=.03), while tumor SUV was significantly correlated with both S-phase percent volume and cell percent volume (P=.02 and .03, respectively). Patlak slope, K(FMAU) and tumor SUV were equivalent with regard to rank correlation analysis, which showed that tumor uptake and trapping of FMAU were correlated with the volume density of dividing cells (P=.0003) rather than nondividing cells (P=.3). Trapping of [(11)C]FMAU correlated with tumor growth rate, as measured by direct tissue analysis with BUdR in a canine brain tumor model, suggesting that [(11)C]FMAU is useful for the imaging of cell proliferation in cancers.

Comparison of SUV and Patlak slope for monitoring of cancer therapy using serial PET scans.

The standardized uptake value (SUV) and the slope of the Patlak plot ( K) have both been proposed as indices to monitor the progress of disease during cancer therapy. Although a good correlation has been reported between SUV and K, they are not equivalent, and may not be equally affected by metabolic changes occurring during disease progression or therapy. We wished to compare changes in tumor SUV with changes in K during serial positron emission tomography (PET) scans for monitoring therapy. Thirteen patients enrolled in a protocol to treat renal cell carcinoma metastases were studied. Serial dynamic fluorodeoxyglucose (FDG) PET scans and computed tomography (CT) and magnetic resonance (MR) scans were performed once prior to treatment, once at 36+/-2 days after the start of treatment, and (in 7/13 subjects, 16/27 lesions) a third time at 92+/-9 days after the start of treatment. This resulted in a total of 33 scans, and 70 tumor Patlak and SUV values (one value for each lesion at each time point). SUV and K were measured over one to four predefined tumors/patient at each time point. The input function was obtained from regions of interest over the heart, combined, if necessary, with late blood samples. Over all tumors and scans, SUV and K correlated well ( r=0.97, P<0.0001). However, change in SUV with treatment over all tumor scan pairs was much less well correlated with the corresponding change in K ( r=0.73, P<0.0001). The absolute difference in % change was outside the 95% confidence limits expected from previous variability studies in 6 of 43 pairs of tumor scans, and greater than 50% in 2 of 43 tumor scan pairs. In four of the six cases, the two indices predicted opposing therapeutic outcomes. Similar results were obtained for SUV normalized by body weight or body surface area and for SUVs using mean or maximum count. Changes in CT and MR tumor cross-product dimensions correlated poorly with each other ( r=0.47, P=NS), and so could not be used to determine the "correct" PET index. Absolute values of SUV and K correlated well over the patient population. However, when monitoring individual patient therapy serially, large differences in the % changes in the two indices were occasionally found, sometimes sufficient to produce opposing conclusions regarding the progression of disease.