Showing papers on "Parametric Image published in 2016"

Parametric Object Motion from Blur

Abstract: Motion blur can adversely affect a number of vision tasks, hence it is generally considered a nuisance. We instead treat motion blur as a useful signal that allows to compute the motion of objects from a single image. Drawing on the success of joint segmentation and parametric motion models in the context of optical flow estimation, we propose a parametric object motion model combined with a segmentation mask to exploit localized, non-uniform motion blur. Our parametric image formation model is differentiable w.r.t. the motion parameters, which enables us to generalize marginal-likelihood techniques from uniform blind deblurring to localized, non-uniform blur. A two-stage pipeline, first in derivative space and then in image space, allows to estimate both parametric object motion as well as a motion segmentation from a single image alone. Our experiments demonstrate its ability to cope with very challenging cases of object motion blur.

Dynamic PET simulator via tomographic emission projection for kinetic modeling and parametric image studies

Abstract: Purpose: To develop and evaluate a fast and simple tool called dpetstep (Dynamic PET Simulator of Tracers via Emission Projection), for dynamic PET simulations as an alternative to Monte Carlo (MC), useful for educational purposes and evaluation of the effects of the clinical environment, postprocessing choices, etc., on dynamic and parametric images. Methods: The tool was developed in matlab using both new and previously reported modules of petstep (PET Simulator of Tracers via Emission Projection). Time activity curves are generated for each voxel of the input parametric image, whereby effects of imaging system blurring, counting noise, scatters, randoms, and attenuation are simulated for each frame. Each frame is then reconstructed into images according to the user specified method, settings, and corrections. Reconstructed images were compared to MC data, and simple Gaussian noised time activity curves (GAUSS). Results: dpetstep was 8000 times faster than MC. Dynamic images from dpetstep had a root mean square error that was within 4% on average of that of MC images, whereas the GAUSS images were within 11%. The average bias in dpetstep and MC images was the same, while GAUSS differed by 3% points. Noise profiles in dpetstep images conformed well to MC images, confirmed visually by scatter plot histograms, and statistically by tumor region of interest histogram comparisons that showed no significant differences (p < 0.01). Compared to GAUSS, dpetstep images and noise properties agreed better with MC. Conclusions: The authors have developed a fast and easy one-stop solution for simulations of dynamic PET and parametric images, and demonstrated that it generates both images and subsequent parametric images with very similar noise properties to those of MC images, in a fraction of the time. They believe dpetstep to be very useful for generating fast, simple, and realistic results, however since it uses simple scatter and random models it may not be suitable for studies investigating these phenomena. dpetstep can be downloaded free of cost from https://github.com/CRossSchmidtlein/dPETSTEP.

Formation of parametric images using mixed-effects models: a feasibility study.

Abstract: Mixed-effects models have been widely used in the analysis of longitudinal data. By presenting the parameters as a combination of fixed effects and random effects, mixed-effects models incorporating both within- and between-subject variations are capable of improving parameter estimation. In this work, we demonstrate the feasibility of using a non-linear mixed-effects (NLME) approach for generating parametric images from medical imaging data of a single study. By assuming that all voxels in the image are independent, we used simulation and animal data to evaluate whether NLME can improve the voxel-wise parameter estimation. For testing purposes, intravoxel incoherent motion (IVIM) diffusion parameters including perfusion fraction, pseudo-diffusion coefficient and true diffusion coefficient were estimated using diffusion-weighted MR images and NLME through fitting the IVIM model. The conventional method of non-linear least squares (NLLS) was used as the standard approach for comparison of the resulted parametric images. In the simulated data, NLME provides more accurate and precise estimates of diffusion parameters compared with NLLS. Similarly, we found that NLME has the ability to improve the signal-to-noise ratio of parametric images obtained from rat brain data. These data have shown that it is feasible to apply NLME in parametric image generation, and the parametric image quality can be accordingly improved with the use of NLME. With the flexibility to be adapted to other models or modalities, NLME may become a useful tool to improve the parametric image quality in the future. Copyright © 2015 John Wiley & Sons, Ltd.

Impact of time-of-flight on indirect 3D and direct 4D parametric image reconstruction in the presence of inconsistent dynamic PET data

Abstract: Kinetic parameter estimation in dynamic PET suffers from reduced accuracy and precision when parametric maps are estimated using kinetic modelling following image reconstruction of the dynamic data. Direct approaches to parameter estimation attempt to directly estimate the kinetic parameters from the measured dynamic data within a unified framework. Such image reconstruction methods have been shown to generate parametric maps of improved precision and accuracy in dynamic PET. However, due to the interleaving between the tomographic and kinetic modelling steps, any tomographic or kinetic modelling errors in certain regions or frames, tend to spatially or temporally propagate. This results in biased kinetic parameters and thus limits the benefits of such direct methods. Kinetic modelling errors originate from the inability to construct a common single kinetic model for the entire field-of-view, and such errors in erroneously modelled regions could spatially propagate. Adaptive models have been used within 4D image reconstruction to mitigate the problem, though they are complex and difficult to optimize. Tomographic errors in dynamic imaging on the other hand, can originate from involuntary patient motion between dynamic frames, as well as from emission/transmission mismatch. Motion correction schemes can be used, however, if residual errors exist or motion correction is not included in the study protocol, errors in the affected dynamic frames could potentially propagate either temporally, to other frames during the kinetic modelling step or spatially, during the tomographic step. In this work, we demonstrate a new strategy to minimize such error propagation in direct 4D image reconstruction, focusing on the tomographic step rather than the kinetic modelling step, by incorporating time-of-flight (TOF) within a direct 4D reconstruction framework. Using ever improving TOF resolutions (580 ps, 440 ps, 300 ps and 160 ps), we demonstrate that direct 4D TOF image reconstruction can substantially prevent kinetic parameter error propagation either from erroneous kinetic modelling, inter-frame motion or emission/transmission mismatch. Furthermore, we demonstrate the benefits of TOF in parameter estimation when conventional post-reconstruction (3D) methods are used and compare the potential improvements to direct 4D methods. Further improvements could possibly be achieved in the future by combining TOF direct 4D image reconstruction with adaptive kinetic models and inter-frame motion correction schemes.

Theoretical Analysis of Penalized Maximum-Likelihood Patlak Parametric Image Reconstruction in Dynamic PET for Lesion Detection

Abstract: Detecting cancerous lesions is a major clinical application of emission tomography. In a previous work, we studied penalized maximum-likelihood (PML) image reconstruction for lesion detection in static PET. Here we extend our theoretical analysis of static PET reconstruction to dynamic PET. We study both the conventional indirect reconstruction and direct reconstruction for Patlak parametric image estimation. In indirect reconstruction, Patlak parametric images are generated by first reconstructing a sequence of dynamic PET images, and then performing Patlak analysis on the time activity curves (TACs) pixel-by-pixel. In direct reconstruction, Patlak parametric images are estimated directly from raw sinogram data by incorporating the Patlak model into the image reconstruction procedure. PML reconstruction is used in both the indirect and direct reconstruction methods. We use a channelized Hotelling observer (CHO) to assess lesion detectability in Patlak parametric images. Simplified expressions for evaluating the lesion detectability have been derived and applied to the selection of the regularization parameter value to maximize detection performance. The proposed method is validated using computer-based Monte Carlo simulations. Good agreements between the theoretical predictions and the Monte Carlo results are observed. Both theoretical predictions and Monte Carlo simulation results show the benefit of the indirect and direct methods under optimized regularization parameters in dynamic PET reconstruction for lesion detection, when compared with the conventional static PET reconstruction.

GIP: Generic Image Prior for No Reference Image Quality Assessment

Abstract: No reference image quality assessment (NR-IQA) has attracted great attention due to the increasing demand in developing perceptually friendly applications. The crucial challenge of this task is how to accurately measure the naturalness of an image. In this paper, we propose a novel parametric image representation which is derived from the generic image prior (GIP). More specifically, we utilize the classic fields of experts model to capture the prior distribution of an image with respect to a random field, which is learned from a great deal of natural images. Then, the parameters in modeling this prior distribution are used as the quality-relevant image feature, which is represented by a simple two-dimension vector. Experimental results show that the proposed method achieves competitive quality prediction accuracy in comparison with the state-of-the-art NR-IQA algorithms at the expense of much less memory usage and computational complexity.

Multiscale Nakagami parametric imaging for improved liver tumor localization

Abstract: Effective ultrasound tissue characterization is usually hindered by complex tissue structures. The interlacing of speckle patterns complicates the correct estimation of backscatter distribution parameters. Nakagami parametric imaging based on localized shape parameter mapping can model different backscattering conditions. However, performance of the constructed Nakagami image depends on the sensitivity of the estimation method to the backscattered statistics and scale of analysis. Using a fixed focal region of interest in estimating the Nakagami parametric image would increase estimation variance. In this work, localized Nakagami parameters are estimated adaptively by means of maximum likelihood estimation on a multiscale basis. The varying size kernel integrates the goodness-of-fit of the backscattering distribution parameters at multiple scales for more stable parameter estimation. Results show improved quantitative visualization of changes in tissue specular reflections, suggesting a potential approach for improving tumor localization in low contrast ultrasound images.

Parametric image segmentation of humans with structural shape priors

Abstract: The figure-ground segmentation of humans in images captured in natural environments is an outstanding open problem due to the presence of complex backgrounds, articulation, varying body proportions, partial views and viewpoint changes. In this work we propose class-specific segmentation models that leverage parametric max-flow image segmentation and a large dataset of human shapes. Our contributions are as follows: (1) formulation of a sub-modular energy model that combines class-specific structural constraints and data-driven shape priors, within a parametric max-flow optimization methodology that systematically computes all breakpoints of the model in polynomial time; (2) design of a data-driven class-specific fusion methodology, based on matching against a large training set of exemplar human shapes (100,000 in our experiments), that allows the shape prior to be constructed on-the-fly, for arbitrary viewpoints and partial views.

Novel quantitative whole-body parametric PET imaging utilizing multiple clustering realizations

Abstract: Whole-body parametric PET imaging along with Patlak graphical analysis has the potential to provide improved diagnosis. However, a voxel-based fitting approach for a short dynamic scan protocol results in high statistical noise in the parametric images. The objective of our study is to present the framework of a novel multiple clustering realizations (MCR) method for estimating parametric images with improved image quality. The method relies primarily on using standard k-means clustering for segmenting the time-activity curves within the whole-body volume. In addition, in order to obtain improved accuracy without increasing noise, multiple realizations of clustering were performed. During each realization, cluster centers were selected from a unique ordered set of time-activity curves within the whole body volume. All the remaining data were classified into the cluster centers based on minimum Eucledian distance measure. Patlak analysis was performed on the cluster average to form the slope and intercept images. Parametric images thus obtained for all realizations were averaged. An XCAT phantom based simulations for the torso were performed using dynamic time-activity curves to model FDG uptake. Five dynamic images each representing 1 min scan time with 7 min intervals were created starting 60 minutes post injection. In addition, 5 whole-body dynamic FDG patient datasets with image-derived blood input function and whole-body dynamic data measurements were also used. All dynamic data were reconstructed using OSEM applying corrections for image-degrading factors. Slope and intercept parametric images were obtained for the voxel-fitting and MCR method. Noise in a liver region of interest increased as a function of the number of clusters for the simulated data. On the other hand, bias decreased with increasing number of clusters. However, as number of clustering realizations increased, noise reduced and K i estimates stabilized. The parametric images obtained with MCR method showed better image quality compared to voxel-based fitting method for the patient and simulated datasets. Multiple clustering realizations method has the potential to provide improved parametric image quality for short scan whole-body parametric PET imaging.

A Flexible 2D-3D Parametric Image Registration Algorithm for Cardiac MRI

Abstract: We propose a mathematical formulation aimed at parametric intensity-based registration of a deformed 3D volume to a 2D slice. The approach is flexible and can accommodate various regularization schemes, similarity measures, and optimizers. We evaluate the framework on 2D-3D registration experiments of in vivo cardiac magnetic resonance imaging (MRI) aimed at image-guided surgery applications that use of real-time MRI as a visualization tool. An affine transformation is used to demonstrate this parametric model. Target registration error, Jaccard and Dice indices are used to validate the algorithm and demonstrate the accuracy of the registration scheme on both simulated and clinical data.

Image construction with multiple clustering realizations

Abstract: A method includes overlaying a grid on a set of dynamic PET, SPECT, CT or MR data, so as to define a set of voxels defining a plurality of cluster seeds; extracting a respective time activity curve (TAC) for dynamic PET or SPECT data or time varying signals in the case of dynamic CT or MR data, for each voxel based on the data; selecting a subset of the cluster seeds defined by the grid as initial cluster centroids of a set of clusters; assigning each TAC to a respective cluster in the set of clusters; computing a respective average TAC of each cluster; generating a parametric image based on the respective average TACs for the clusters; repeating the overlaying, determining, selecting, assigning, computing, and generating; and averaging the generated parametric images

Time of flight parametric image reconstruction from variable random fraction dynamic PET data

Abstract: Dynamic PET imaging allows the time course of the activity distribution to be measured and modelled, therefore estimating parametric images of micro- or macro-parameters. Due to the need for increased temporal sampling, frames of low statistics are often reconstructed, leading to noisy dynamic data and subsequently to kinetic parameters of reduced precision and accuracy. TOF image reconstruction can substantially improve upon the kinetic parameter SNR. However, variable contributions to TOF variance reduction between true and random events, owing to their spatial distribution, results in SNR improvements depending on the random fraction and leading to variable TOF gains amongst temporal frames within a dynamic study. Such variable gains between early/late frames (high/low random fractions) are also expected to be more pronounced at increasing injected doses. Therefore, we hypothesize that certain kinetic parameters receive differential improvements depending upon the part of the time-activity curve they are estimated from. Using simulated dynamic [15O]H 2 O datasets at ever increasing doses and random fractions, we investigate the variable TOF gain in dynamic imaging and its effect on the kinetic parameters. Data are presented at improving TOF resolutions and using both indirect and direct methods to kinetic parameter estimation. Initial results suggest that kinetic parameter TOF gain is highly variable (increasing) at ever increasing injected doses, but such variation is different for each parameter based on the part of the dynamic data it is derived from, owing to the variable TOF gain within dynamic frames.