An important task when processing dynamic PET images is to identify the time-activity curves (TACs) of the pure tissues, along with their corresponding spatial proportions. This step, often referred to as unmixing or factor analysis, is based on a loss function which measures the discrepancy between the observed data and the model. This loss function should be chosen according to the statistical properties of the noise, which is in this case hard to characterize. Indeed, while dynamic PET images results from a decay process that can be statistically described by a Poisson distribution, acquisition and post-filtering reconstruction drastically change the nature of the noise. In the literature dedicated to factor analysis of dynamic PET images, a common and underlying assumption consists in assuming that the dynamic PET images are corrupted by an additive Gaussian or by a Poisson noise. These assumptions lead to the choice of the squared Euclidian distance and the Kullback-Leibler divergence. We propose here to consider the β-divergence, which is able to encompass a wide family of divergence measures corresponding to various noise distributions. This loss function is incorporated into three different factor models and evaluated using four sets of synthetic data.

Factor analysis has proven to be a relevant tool for extracting tissue time-activity curves (TACs) in dynamic PET images, since it allows for an unsupervised analysis of the data. Reliable and interpretable results are possible only if it is considered with respect to suitable noise statistics. However, the noise in reconstructed dynamic PET images is very difficult to characterize, despite the Poissonian nature of the count rates. Rather than explicitly modeling the noise distribution, this paper proposes to study the relevance of several divergence measures to be used within a factor analysis framework. To this end, the β-divergence, widely used in other applicative domains, is considered to design the data-fitting term involved in three different factor models. The performances of the resulting algorithms are evaluated for different values of β, in a range covering Gaussian, Poissonian, and Gamma-distributed noises. The results obtained on two different types of synthetic images and one real image show the interest of applying non-standard values of β to improve the factor analysis.

In PET image reconstruction, it would be useful to obtain the entire posterior probability distribution of the image, because it allows for both estimating image intensity and assessing the uncertainty of the estimation, thus leading to more reliable interpretation. We propose a new entirely probabilistic model: the prior is a distribution over possible smooth regions (distance-driven Chinese restaurant process), and the posterior distribution is estimated using a Gibbs Markov chain Monte Carlo sampler. Data from other modalities (here one or several MR images) are introduced into the model as additional observed data, providing side information about likely smooth regions in the image. The reconstructed image is the posterior mean, and the uncertainty is presented as an image of the size of 95% posterior intervals. The reconstruction was compared with the maximum-likelihood expectation-maximization and OSEM algorithms, with and without post-smoothing, and with a penalized ML or MAP method that also uses additional images from other modalities. Qualitative and quantitative tests were performed on realistic simulated data with statistical replicates and on several clinical examinations presenting pathologies. The proposed method presents appealing properties in terms of obtained bias, variance, spatial regularization, and use of multimodal data, and produces, in addition, potentially valuable uncertainty information.

Positron emission tomography data are typically reconstructed with maximum likelihood expectation maximization (MLEM). However, MLEM suffers from positive bias due to the non-negativity constraint. This is particularly problematic for tracer kinetic modeling. Two reconstruction methods with bias reduction properties that do not use strict Poisson optimization are presented and compared to each other, to filtered backprojection (FBP), and to MLEM. The first method is an extension of NEGML, where the Poisson distribution is replaced by a Gaussian distribution for low count data points. The transition point between the Gaussian and the Poisson regime is a parameter of the model. The second method is a simplification of ABML. ABML has a lower and upper bound for the reconstructed image whereas AML has the upper bound set to infinity. AML uses a negative lower bound to obtain bias reduction properties. Different choices of the lower bound are studied. The parameter of both algorithms determines the effectiveness of the bias reduction and should be chosen large enough to ensure bias-free images. This means that both algorithms become more similar to least squares algorithms, which turned out to be necessary to obtain bias-free reconstructions. This comes at the cost of increased variance. Nevertheless, NEGML and AML have lower variance than FBP. Furthermore, randoms handling has a large influence on the bias. Reconstruction with smoothed randoms results in lower bias compared to reconstruction with unsmoothed randoms or randoms precorrected data. However, NEGML and AML yield both bias-free images for large values of their parameter.

In this paper, we extend the gradient vector flow field for robust variational segmentation of vector-valued images. Rather than using scalar edge information, we define a vectorial edge map derived from a weighted local structure tensor of the image that enables the diffusion of the gradient vectors in accurate directions through the 4D gradient vector flow equation. To reduce the contribution of noise in the structure tensor, image channels are weighted according to a blind estimator of contrast. The method is applied to biological volume delineation in dynamic PET imaging, and validated on realistic Monte Carlo simulations of numerical phantoms as well as on real images.

We propose a nonparametric and Bayesian method for reconstructing dynamic Positron Emission Tomography (PET) images from clinical data. PET is a nuclear medicine imaging modality that uses molecules labeled with a positron emitting radionuclide. It is then possible to image in vivo molecular interactions of biological processes. Our approach is non-parametric in the sense that the image representing the 4D (3D+t) activity distribution is viewed as a probability density on R3× R+and inferred directly from the data, without any prior space or time discretization. Being nonparametric, we do not pre-assume any particular functional form for this space-time distribution. Formulating the nonparametric problem in the Bayesian framework allows to characterize the entire 4D distribution of the unknown. Furthermore, this framework allows to access directly to the reconstruction error. The ability of the proposed model is assessed using data from clinical studies and we evaluate its performance against the conventional independent time-frame reconstruction approach using the maximum likelihood algorithm (ML-EM).

18F-fluorodeoxyglucose positron emission tomography (18FDG PET) has become an essential technique in oncology. Accurate segmentation and uptake quantification are crucial in order to enable objective follow-up, the optimization of radiotherapy planning, and therapeutic evaluation. We have designed and evaluated a new, nearly automatic and operator-independent segmentation approach. This incorporated possibility theory, in order to take into account the uncertainty and inaccuracy inherent in the image. The approach remained independent of PET facilities since it did not require any preliminary calibration. Good results were obtained from phantom images [percent error =18.38% (mean) ± 9.72% (standard deviation)]. Results on simulated and anatomopathological data sets were quantified using different similarity measures and showed the method was efficient (simulated images: Dice index =82.18% ± 13.53% for SUV =2.5 ). The approach could, therefore, be an efficient and robust tool for uptake volume segmentation, and lead to new indicators for measuring volume of interest activity.