ParaPET: non-invasive deep learning method for direct parametric brain PET reconstruction using histoimages.

BACKGROUND: The indirect method for generating parametric images in positron emission tomography (PET) involves the acquisition and reconstruction of dynamic images and temporal modelling of tissue activity given a measured arterial input function. This approach is not robust, as noise in each dynamic image leads to a degradation in parameter estimation. Direct methods incorporate into the image reconstruction step both the kinetic and noise models, leading to improved parametric images. These methods require extensive computational time and large computing resources. Machine learning methods have demonstrated significant potential in overcoming these challenges. But they are limited by the requirement of a paired training dataset. A further challenge within the existing framework is the use of state-of-the-art arterial input function estimation via temporal arterial blood sampling, which is an invasive procedure, or an additional magnetic resonance imaging (MRI) scan for selecting a region where arterial blood signal can be measured from the PET image. We propose a novel machine learning approach for reconstructing high-quality parametric brain images from histoimages produced from time-of-flight PET data without requiring invasive arterial sampling, an MRI scan, or paired training data from standard field-of-view scanners. RESULT: The proposed is tested on a simulated phantom and five oncological subjects undergoing an 18F-FDG-PET scan of the brain using Siemens Biograph Vision Quadra. Kinetic parameters set in the brain phantom correlated strongly with the estimated parameters (K1, k2 and k3, Pearson correlation coefficient of 0.91, 0.92 and 0.93) and a mean squared error of less than 0.0004. In addition, our method significantly outperforms (p < 0.05, paired t-test) the conventional nonlinear least squares method in terms of contrast-to-noise ratio. At last, the proposed method was found to be 37% faster than the conventional method. CONCLUSION: We proposed a direct non-invasive DL-based reconstruction method and produced high-quality parametric maps of the brain. The use of histoimages holds promising potential for enhancing the estimation of parametric images, an area that has not been extensively explored thus far. The proposed method can be applied to subject-specific dynamic PET data alone.

Unsupervised PET logan parametric image estimation using conditional deep image prior.

Recently, deep learning-based denoising methods have been gradually used for PET images denoising and have shown great achievements. Among these methods, one interesting framework is conditional deep image prior (CDIP) which is an unsupervised method that does not need prior training or a large number of training pairs. In this work, we combined CDIP with Logan parametric image estimation to generate high-quality parametric images. In our method, the kinetic model is the Logan reference tissue model that can avoid arterial sampling. The neural network was utilized to represent the images of Logan slope and intercept. The patient's computed tomography (CT) image or magnetic resonance (MR) image was used as the network input to provide anatomical information. The optimization function was constructed and solved by the alternating direction method of multipliers (ADMM) algorithm. Both simulation and clinical patient datasets demonstrated that the proposed method could generate parametric images with more detailed structures. Quantification results showed that the proposed method results had higher contrast-to-noise (CNR) improvement ratios (PET/CT datasets: 62.25%±29.93%; striatum of brain PET datasets : 129.51%±32.13%, thalamus of brain PET datasets: 128.24%±31.18%) than Gaussian filtered results (PET/CT datasets: 23.33%±18.63%; striatum of brain PET datasets: 74.71%±8.71%, thalamus of brain PET datasets: 73.02%±9.34%) and nonlocal mean (NLM) denoised results (PET/CT datasets: 37.55%±26.56%; striatum of brain PET datasets: 100.89%±16.13%, thalamus of brain PET datasets: 103.59%±16.37%).

Correlated-Weighted Statistically Modeled Contourlet and Curvelet Coefficient Image-Based Breast Tumor Classification Using Deep Learning.

Deep learning-based automatic classification of breast tumors using parametric imaging techniques from ultrasound (US) B-mode images is still an exciting research area. The Rician inverse Gaussian (RiIG) distribution is currently emerging as an appropriate example of statistical modeling. This study presents a new approach of correlated-weighted contourlet-transformed RiIG (CWCtr-RiIG) and curvelet-transformed RiIG (CWCrv-RiIG) image-based deep convolutional neural network (CNN) architecture for breast tumor classification from B-mode ultrasound images. A comparative study with other statistical models, such as Nakagami and normal inverse Gaussian (NIG) distributions, is also experienced here. The weighted entitled here is for weighting the contourlet and curvelet sub-band coefficient images by correlation with their corresponding RiIG statistically modeled images. By taking into account three freely accessible datasets (Mendeley, UDIAT, and BUSI), it is demonstrated that the proposed approach can provide more than 98 percent accuracy, sensitivity, specificity, NPV, and PPV values using the CWCtr-RiIG images. On the same datasets, the suggested method offers superior classification performance to several other existing strategies.

Generation of synthetic PET images of synaptic density and amyloid from 18 F-FDG images using deep learning.

PURPOSE: Positron emission tomography (PET) imaging with various tracers is increasingly used in Alzheimer's disease (AD) studies. However, access to PET scans using new or less-available tracers with sophisticated synthesis and short half-life isotopes may be very limited. Therefore, it is of great significance and interest in AD research to assess the feasibility of generating synthetic PET images of less-available tracers from the PET image of another common tracer, in particular 18 F-FDG. METHODS: We implemented advanced deep learning methods using the U-Net model to predict 11 C-UCB-J PET images of synaptic vesicle protein 2A (SV2A), a surrogate of synaptic density, from 18 F-FDG PET data. Dynamic 18 F-FDG and 11 C-UCB-J scans were performed in 21 participants with normal cognition (CN) and 33 participants with Alzheimer's disease (AD). Cerebellum was used as the reference region for both tracers. For 11 C-UCB-J image prediction, four network models were trained and tested, which included 1) 18 F-FDG SUV ratio (SUVR) to 11 C-UCB-J SUVR, 2) 18 F-FDG Ki ratio to 11 C-UCB-J SUVR, 3) 18 F-FDG SUVR to 11 C-UCB-J distribution volume ratio (DVR), and 4) 18 F-FDG Ki ratio to 11 C-UCB-J DVR. The normalized root mean square error (NRMSE), structure similarity index (SSIM), and Pearson's correlation coefficient were calculated for evaluating the overall image prediction accuracy. Mean bias of various ROIs in the brain and correlation plots between predicted images and true images were calculated for ROI-based prediction accuracy. Following a similar training and evaluation strategy, 18 F-FDG SUVR to 11 C-PiB SUVR network was also trained and tested for 11 C-PiB static image prediction. RESULTS: The results showed that all four network models obtained satisfactory 11 C-UCB-J static and parametric images. For 11 C-UCB-J SUVR prediction, the mean ROI bias was -0.3% ± 7.4% for the AD group and -0.5% ± 7.3% for the CN group with 18 F-FDG SUVR as the input, -0.7% ± 8.1% for the AD group, and -1.3% ± 7.0% for the CN group with 18 F-FDG Ki ratio as the input. For 11 C-UCB-J DVR prediction, the mean ROI bias was -1.3% ± 7.5% for the AD group and -2.0% ± 6.9% for the CN group with 18 F-FDG SUVR as the input, -0.7% ± 9.0% for the AD group, and -1.7% ± 7.8% for the CN group with 18 F-FDG Ki ratio as the input. For 11 C-PiB SUVR image prediction, which appears to be a more challenging task, the incorporation of additional diagnostic information into the network is needed to control the bias below 5% for most ROIs. CONCLUSIONS: It is feasible to use 3D U-Net-based methods to generate synthetic 11 C-UCB-J PET images from 18 F-FDG images with reasonable prediction accuracy. It is also possible to predict 11 C-PiB SUVR images from 18 F-FDG images, though the incorporation of additional non-imaging information is needed.

Region- and voxel-based quantification in human brain of [18F]LSN3316612, a radioligand for O-GlcNAcase.

BACKGROUND: Previous studies found that the positron emission tomography (PET) radioligand [18F]LSN3316612 accurately quantified O-GlcNAcase in human brain using a two-tissue compartment model (2TCM). This study sought to assess kinetic model(s) as an alternative to 2TCM for quantifying [18F]LSN3316612 binding, particularly in order to generate good-quality parametric images. METHODS: The current study reanalyzed data from a previous study of 10 healthy volunteers who underwent both test and retest PET scans with [18F]LSN3316612. Kinetic analysis was performed at the region level with 2TCM using 120-min PET data and arterial input function, which was considered as the gold standard. Quantification was then obtained at both the region and voxel levels using Logan plot, Ichise's multilinear analysis-1 (MA1), standard spectral analysis (SA), and impulse response function at 120 min (IRF120). To avoid arterial sampling, a noninvasive relative quantification (standardized uptake value ratio (SUVR)) was also tested using the corpus callosum as a pseudo-reference region. Venous samples were also assessed to see whether they could substitute for arterial ones. RESULTS: Logan and MA1 generated parametric images of good visual quality and their total distribution volume (VT) values at both the region and voxel levels were strongly correlated with 2TCM-derived VT (r = 0.96-0.99) and showed little bias (up to - 8%). SA was more weakly correlated to 2TCM-derived VT (r = 0.93-0.98) and was more biased (~ 16%). IRF120 showed a strong correlation with 2TCM-derived VT (r = 0.96) but generated noisier parametric images. All techniques were comparable to 2TCM in terms of test-retest variability and reliability except IRF120, which gave significantly worse results. Noninvasive SUVR values were not correlated with 2TCM-derived VT, and arteriovenous equilibrium was never reached. CONCLUSIONS: Compared to SA and IRF, Logan and MA1 are more suitable alternatives to 2TCM for quantifying [18F]LSN3316612 and generating good-quality parametric images.

Arrival time parametric imaging of contrast enhanced ultrasound for differential diagnosis of gallbladder lesions / 中国医学影像技术

Objective: To explore the value of arrival time parametric imaging (ATPI) of contrast-enhanced ultrasonography (CEUS) for differential diagnosis of gallbladder lesions. Methods: Data of 25 patients with gallbladder carcinoma and 22 with benign gallbladder lesions who received CEUS and cholecystectomy were reviewed. ATPI system was used to analyze CEUS imaging. The artery patterns of the lesions on ATPI in artery phase were evaluated. Then quantitative parameters, including arrival time of gallbladder lesions, arrival time of surrounding liver as well as the time difference (△T) were analyzed. Results: The artery patterns were different between benign and malignant gallbladder lesions in artery phase (P0.05). However, △T (the arrival time difference of lesions and the surrounding liver) of benign (-0.21±1.37)s and malignant (-2.69±1.37)s lesions were statistically significant (P<0.001).Taken -1.05 s as the cut off value of △T, the sensitivity, specificity, positive predictive value, negative predictive and value accuracy in differentiating gallbladder carcinoma from benign lesions was 81.80%, 92.00%, 85.20%, 90.00% and 87.20%, respectively. Conclusion: Artery patterns and parameters of ATPI of CEUS are helpful to differential diagnosis of benign and malignant gallbladder lesions.

[Quantitative methods in sonography]. / Quantitative Verfahren in der Sonographie.

B­mode and color Doppler ultrasound are standard radiological methods to quantify tissue echo texture and tissue perfusion. Microstructure and composition of tissue influence echo texture parameters and acoustic parameters, such as speed of sound, attenuation and backscatter and quantitative color Doppler image parameters are influenced by the hemodynamics in depictable vessels. Dynamic contrast-enhanced ultrasound enables quantification of tissue perfusion and ultrasound elastography assists in assessing tissue stiffness. B­mode texture analysis, analysis of high-frequency echo signals and quantitative color Doppler image analysis are able to contribute to the assessment of tissue microstructure but have so far not been implemented clinically due to their complexity. Dynamic contrast-enhanced ultrasound and ultrasound elastography have proven to be robust under clinical conditions.

Comparative evaluation of the algorithms for parametric mapping of the novel myocardial PET imaging agent 18F-FPTP.

OBJECTIVE: (18F-fluoropentyl)triphenylphosphonium salt (18F-FPTP) is a new promising myocardial PET imaging tracer. It shows high accumulation in cardiomyocytes and rapid clearance from liver. We performed compartmental analysis of 18F-FPTP PET images in rat and evaluated two linear analyses: linear least-squares (LLS) and a basis function method (BFM) for generating parametric images. The minimum dynamic scan duration for kinetic analysis was also investigated and computer simulation undertaken. METHODS: 18F-FPTP dynamic PET (18 min) and CT images were acquired from rats with myocardial infarction (MI) (n = 12). Regions of interest (ROIs) were on the left ventricle, normal myocardium, and MI region. Two-compartment (K 1 and k 2; 2C2P) and three-compartment (K 1-k 3; 3C3P) models with irreversible uptake were compared for goodness-of-fit. Partial volume and spillover correction terms (V a and α = 1 - V a ) were also incorporated. LLS and BFM were applied to ROI- and voxel-based kinetic parameter estimations. Results were compared with the standard ROI-based nonlinear least-squares (NLS) results of the corresponding compartment model. A simulation explored statistical properties of the estimation methods. RESULTS: The 2C2P model was most suitable for describing 18F-FPTP kinetics. Average K 1, k 2, and V a values were, respectively, 6.8 (ml/min/g), 1.1 (min-1), and 0.44 in normal myocardium and 1.4 (ml/min/g), 1.1 (min-1), and 0.32, in MI tissue. Ten minutes of data was sufficient for the estimation. LLS and BFM estimations correlated well with NLS values for the ROI level (K 1: y = 1.06x + 0.13, r 2  = 0.96 and y = 1.13x + 0.08, r 2  = 0.97) and voxel level (K 1: y = 1.22x - 0.30, r 2  = 0.90 and y = 1.26x + 0.00, r 2  = 0.92). Regional distribution of kinetic parametric images (αK 1, K 1, k 2, V a) was physiologically relevant. LLS and BFM showed more robust characteristics than NLS in the simulation. CONCLUSIONS: Fast kinetics and highly specific uptake of 18F-FPTP by myocardium enabled quantitative analysis with the 2C2P model using only the initial 10 min of data. LLS and BFM were feasible for estimating voxel-wise parameters. These two methods will be useful for quantitative evaluation of 18F-FPTP distribution in myocardium and in further studies with different conditions, disease models, and species.

Simplified estimation of binding parameters based on image-derived reference tissue models for dopamine transporter bindings in non-human primates using [18F]FE-PE2I and PET.

The aim of this study on dopamine transporter binding by [18F]FE-PE2I and PET was to describe an image-derived approach using reference tissue models: the Logan DVR approach and simplified reference tissue model (SRTM), the features of which were simple to operate and precise in the measurements. Using the approach, the authors sought to obtain binding images and parameters. [18F]FE-PE2I and dynamic PET as well as an MRI was performed on three rhesus monkeys, and metabolite corrected arterial plasma inputs were obtained. After co-registering of PET to MR images, both image sets were resliced. The time-activity curve of the cerebellum was used as indirect input, and binding parametric images were computed voxel-by-voxel. Voxel-wise linear calculations were used for the Logan DVR approach, and nonlinear least squares fittings for the SRTM. To determine the best linear regression in the Logan DVR approach, the distribution volume ratio was obtained using the optimal starting frame analysis. The obtained binding parameters were compared with those obtained by the other independent ROI-based numerical approaches: two-tissue compartment model (2TCM), Logan DVR approach and SRTM using PMOD software. Binding potentials (BP) obtained by the present approach agreed well with those obtained by ROI-based numerical approaches, although reference tissue models tended to underestimate the BP value than 2TCM. Image-derived Logan approach provided a low-noise image, the computation time was short, and the error in the optimal starting frame analysis was small. The present approach provides a high-quality binding parametric image and reliable parameter value easily.

Parameter estimation of perfusion models in dynamic contrast-enhanced imaging: a unified framework for model comparison.

Patients follow-up in oncology is generally performed through the acquisition of dynamic sequences of contrast-enhanced images. Estimating parameters of appropriate models of contrast intake diffusion through tissues should help characterizing the tumour physiology. However, several models have been developed and no consensus exists on their clinical use. In this paper, we propose a unified framework to analyse models of perfusion and estimate their parameters in order to obtain reliable and relevant parametric images. After defining the biological context and the general form of perfusion models, we propose a methodological framework for model assessment in the context of parameter estimation from dynamic imaging data: global sensitivity analysis, structural and practical identifiability analysis, parameter estimation and model comparison. Then, we apply our methodology to five of the most widely used compartment models (Tofts model, extended Tofts model, two-compartment model, tissue-homogeneity model and distributed-parameters model) and illustrate the results by analysing the behaviour of these models when applied to data acquired on five patients with abdominal tumours.

Impact of parametric imaging on contrast-enhanced ultrasound of breast cancer.

PURPOSE: To prospectively evaluate the usefulness of contrast-enhanced ultrasound (CEUS) using parametric imaging for breast cancer in a multicenter study. METHODS: A total of 65 patients with breast cancer were included in this study. CEUS was performed, and still images on peak time (S), accumulated images (A) and parametric images (P) were generated from the raw data. Four blind reviewers ranked the best visible images as first place, and determined second and third place consecutively. We compared the average ranking of each image. The maximal diameter of the tumor determined on ultrasonography and MRI was compared with the corresponding pathological maximal diameter for 48 of the 65 patients. The correlation between the diameter determined by two experts and two beginners was analyzed. RESULTS: The average rank of visibility was as follows: P, 1.44; A, 2.04; and S, 2.52. The correlation between each image and the pathology was as follows: P, r = 0.664; A, r = 0.630; S, r = 0.717; and MRI, r = 0.936. There were no significant differences among the correlation between the experts and beginners in each image. CONCLUSIONS: The use of parametric imaging improves the visibility of CEUS. The maximal diameter of the tumor determined on CEUS correlates substantially with the pathology.

Separable effects of inversion and contrast-reversal on face detection thresholds and response functions: a sweep VEP study.

The human brain rapidly detects faces in the visual environment. We recently presented a sweep visual evoked potential approach to objectively define face detection thresholds as well as suprathreshold response functions (Ales, Farzin, Rossion, & Norcia, 2012). Here we determined these parameters are affected by orientation (upright vs. inverted) and contrast polarity (positive vs. negative), two manipulations that disproportionately disrupt the perception of faces relative to other object categories. Face stimuli parametrically increased in visibility through phase-descrambling while alternating with scrambled images at a fixed presentation rate of 3 Hz (6 images/s). The power spectrum and mean luminance of all stimuli were equalized. As a face gradually emerged during a stimulation sequence, EEG responses at 3 Hz appeared at ≈35% phase coherence over right occipito-temporal channels, replicating previous observations. With inversion and contrast-reversal, the 3-Hz amplitude decreased by ≈20%-50% and the face detection threshold increased by ≈30%-60% coherence. Furthermore, while the 3-Hz response emerged abruptly and saturated quickly for normal faces, suggesting a categorical neural response, the response profile for inverted and negative polarity faces was shallower and more linear, indicating gradual and continuously increasing activation of the underlying neural population. These findings demonstrate that inversion and contrast-reversal increase the threshold and modulate the suprathreshold response function of face detection.

Angiographic analysis for phantom simulations of endovascular aneurysm treatments with a new fully retrievable asymmetric flow diverter.

Digital Subtraction Angiography (DSA) is the main diagnostic tool for intracranial aneurysms (IA) flow-diverter (FD) assisted treatment. Based on qualitative contrast flow evaluation, interventionists decide on subsequent steps. We developed a novel fully Retrievable Asymmetric Flow-Diverter (RAFD) which allows controlled deployment, repositioning and detachment achieve optimal flow diversion. The device has a small low porosity or solid region which is placed such that it would achieve maximum aneurysmal in-jet flow deflection with minimum impairment to adjacent vessels. We tested the new RAFD using a flow-loop with an idealized and a patient specific IA phantom in carotid-relevant physiological conditions. We positioned the deflection region at three locations: distally, center and proximally to the aneurysm orifice and analyzed aneurysm dome flow using DSA derived maps for mean transit time (MTT) and bolus arrival times (BAT). Comparison between treated and untreated (control) maps quantified the RAFD positioning effect. Average MTT, related to contrast presence in the aneurysm dome increased, indicating flow decoupling between the aneurysm and parent artery. Maximum effect was observed in the center and proximal position (~75%) of aneurysm models depending on their geometry. BAT maps, correlated well with inflow jet direction and magnitude. Reduction and jet dispersion as high as about 50% was observed for various treatments. We demonstrated the use of DSA data to guide the placement of the RAFD and showed that optimum flow diversion within the aneurysm dome is feasible. This could lead to more effective and a safer IA treatment using FDs.

A non-parametric segmentation methodology for oral videocapillaroscopic images.

We aim to describe a new non-parametric methodology to support the clinician during the diagnostic process of oral videocapillaroscopy to evaluate peripheral microcirculation. Our methodology, mainly based on wavelet analysis and mathematical morphology to preprocess the images, segments them by minimizing the within-class luminosity variance of both capillaries and background. Experiments were carried out on a set of real microphotographs to validate this approach versus handmade segmentations provided by physicians. By using a leave-one-patient-out approach, we pointed out that our methodology is robust, according to precision-recall criteria (average precision and recall are equal to 0.924 and 0.923, respectively) and it acts as a physician in terms of the Jaccard index (mean and standard deviation equal to 0.858 and 0.064, respectively).

Parametric imaging using contrast-enhanced ultrasound with Sonazoid for hepatocellular carcinoma.

PURPOSE: To clarify the usefulness of parametric imaging using contrast-enhanced ultrasound (CE-US) with Sonazoid by comparing parametric images of hepatocellular carcinoma (HCC) with histopathological findings. METHODS: Two patients with HCCs underwent CE-US with Sonazoid before surgical resection. A single focus point was set at the lower margin of the tumor, and a bolus intravenous injection of Sonazoid (0.5 ml) was administered. Images of the ideal scanning plane were displayed in real-time mode for the early vascular phase. We analyzed these images using prototype PC software. The software watches, pixel by pixel, the increase in the intensity due to the inflow of the microbubbles, and displays colors if the intensity becomes larger than a certain threshold. Parametric images were compared with histopathological findings. RESULTS: The level of blood flow in the tumor could be visually evaluated using a single image by expressing the detailed hemodynamics of the tumor in terms of differences in color using a time axis appropriate for each case. CONCLUSIONS: Parametric imaging is a very useful way of facilitating straightforward visualization of the level of blood flow within HCC and the distribution of histopathological findings in single static images.

Parametric imaging of contrast-enhanced ultrasound for hepatocellular carcinoma / 中华超声影像学杂志

Objective To investigate the usefulness of parametric imaging of contrast-enhanced ultrasound(CEUS) in imaging hepatocellular carcinoma (HCC) using dynamic vascular patterns (DVP). Methods Thirty clinically or pathologically proven HCCs that had undergone CEUS were randomly included. SonoLiver CAP sofeware was used to analyze the CEUS images and reconstruct DVP parametric images. Results The rise time, time to peak and mean transit time were (16. 72±11. 07) s, (29. 92±14. 13) s,(115. 03±90. 91)s in HCC versus (26. 59±9. 60) s, (41.67±12. 59) s, (159.26±123. 74) s in the surrounding liver parenchyma (all P <0. 05). The perfusion index was (90. 41±102. 49) % in HCC versus (54. 10±24. 99)% in surrounding liver parenchyma( P = 0.044). DVP curve and DVP parametric image could both be divided into three types:washout,non-washout and cystic type. The percentages of which were 76.7% (23/30), 20.0% (6/30) and 3.3% (1/30) in DVP curves, respectively, and 66.7% (20/30), 30.0% (9/30) and 3.3% (1/30) in DVP parametric images,respectively. Conclusions Parametric image of CEUS could demonstrates the difference of flow perfusion static between HCC and surrounding liver parenchyma dynamically and directly.

Linearized Methods for Quantitative Analysis and Parametric Mapping of Brain PET / 핵의학분자영상

Quantitative analysis of dynamic brain PET data using a tracer kinetic modeling has played important roles in the investigation of functional and molecular basis of various brain diseases. Parametric imaging of the kinetic parameters (voxel-wise representation of the estimated parameters) has several advantages over the conventional approaches using region of interest (ROI). Therefore, several strategies have been suggested to generate the parametric images with a minimal bias and variability in the parameter estimation. In this paper, we will review the several approaches for parametric imaging with linearized methods which include graphical analysis and mulilinear regression analysis.

Multiple Linear Analysis for Generating Parametric Images of Irreversible Radiotracer / 핵의학분자영상

PURPOSE: Biological parameters can be quantified using dynamic PET data with compartment modeling and Nonlinear Least Square (NLS) estimation. However, the generation of parametric images using the NLS is not appropriate because of the initial value problem and excessive computation time. In irreversible model, Patlak graphical analysis (PGA) has been commonly used as an alternative to the NLS method. In PGA, however, the start time (t*, time where linear phase starts) has to be determined. In this study, we suggest a new Multiple Linear Analysis for irreversible radiotracer (MLAIR) to estimate fluoride bone influx rate (Ki). METHODS: [18F]Fluoride dynamic PET scans was acquired for 60 min in three normal mini-pigs. The plasma input curve was derived using blood sampling from the femoral artery. Tissue time-activity curves were measured by drawing region of interests (ROIs) on the femur head, vertebra, and muscle. Parametric images of Ki were generated using MLAIR and PGA methods. RESULT: In ROI analysis, estimated Ki values using MLAIR and PGA method was slightly higher than those of NLS, but the results of MLAIR and PGA were equivalent. Patlak slopes (Ki) were changed with different t* in low uptake region. Compared with PGA, the quality of parametric image was considerably improved using new method. CONCLUSION: The results showed that the MLAIR was efficient and robust method for the generation of Ki parametric image from [18F]Fluoride PET. It will be also a good alternative to PGA for the radiotracers with irreversible three compartment model.

Comparison of Algorithms for Generating Parametric Image of Cerebral Blood Flow Using H215O Positron Emission Tomography / 대한핵의학회잡지

PURPOSE: To obtain regional blood flow and tissue-blood partition coefficient with time-activity curves from H2 (15) O PET, fitting of some parameters in the Kety model is conventionally accomplished by nonlinear least squares (NLS) analysis. However, NLS requires considerable compuation time then is impractical for pixel-by-pixel analysis to generate parametric images of these parameters. In this study, we investigated several fast parameter estimation methods for the parametric image generation and compared their statistical reliability and computational efficiency. MATERIALS AND METHODS: These methods included linear least squres (LLS), linear weighted least squares (LWLS), linear generalized least squares (GLS), linear generalized weighted least squares (GWLS), weighted integration (WI), and model-based clustering method (CAKS). H2 (15) O dynamic brain PET with Poisson noise component was simulated using numerical Zubal brain phantom. Error and bias in the estimation of rCBF and partition coefficient, and computation time in various noise environments was estimated and compared. In addition, parametric images from H2 (15) O dynamic brain PET data performed on 16 healthy volunteers under various physiological conditions was compared to examine the utility of these methods for real human data. RESULTS: These fast algorithms produced parametric images with similar image quality and statistical reliability. When CAKS and LLS methods were used combinedly, computation time was significantly reduced and less than 30 seconds for 128x128x46 images on Pentium III processor. CONCLUSION: Parametric images of rCBF and partition coefficient with good statistical properties can be generated with short computation time which is acceptable in clinical situation.