Methods like PET imaging which use radiotracers that bind to amyloid-β (Aβ) - plaques that accumulate and lead to Alzheimer's disease (AD) - are crucial to detect AD early. However, current semi-quantitative methods that quantify Aβ using the Standardized Uptake Values Ratio (SUVr) cannot distinguish between the specific binding of radiotracers to Aβ and non-specific binding to other targets. In this paper, we propose a novel method to predict non-specific binding from MR images using deep learning by developing a novel conditional Generative Adversarial Network (cGAN) to improve Aβ quantification. PET images with low amyloid load were selected to represent non-specific uptake of the Aβ [11C]PiB radiotracer. Paired MR and PET images with low amyloid load were used for training, with the MRI image as the network input and the PET image as the training label. The cGAN was trained to learn the mapping between the MR image and the non-specific PET image. The generated non-specific PET images were compared to the real PET scans, achieving a mean SUVr difference of 1.90% in the gray matter. The small SUVr difference indicated that we have successfully developed cGAN for predicting the non-specific uptake of PET and that deep learning can be used to estimate the non-specific binding of Aβ PET radiotracer from MR images.

A wide range of medical imaging applications benefits from the availability of realistic ground truth data. In the case of positron emission tomography (PET), ground truth data is crucial to validate processing algorithms and assessing their performances. The design of such ground truth data often relies on Monte-Carlo simulation techniques. Since the creation of a large dataset is not trivial both in terms of computing time and realism, we propose the OSSI-PET database containing 350 simulated [(11)C]Raclopride dynamic scans for rats, created specifically for the Inveon pre-clinical PET scanner. The originality of this database lies on the availability of several groups of scans with controlled biological variations in the striata. Besides, each group consists of a large number of realizations (i.e., noise replicates). We present the construction methodology of this database using rat pharmacokinetic and anatomical models. A first application using the OSSI-PET database is presented. Several commonly used reconstruction techniques were compared in terms of image quality, accuracy and variability of the activity estimates and of the computed kinetic parameters. The results showed that OP-OSEM3D iterative reconstruction method outperformed the other tested methods. Analytical methods such as FBP2D and 3DRP also produced satisfactory results. However, FORE followed by OSEM2D reconstructions should be avoided. Beyond the illustration of the potential of the database, this application will help scientists to understand the different sources of noise and bias that can occur at the different steps in the processing and will be very useful for choosing appropriate reconstruction methods and parameters.

In emission tomography, image reconstruction and therefore also tracer development and diagnosis may benefit from the use of anatomical side information obtained with other imaging modalities in the same subject, as it helps to correct for the partial volume effect. One way to implement this, is to use the anatomical image for defining the a priori distribution in a maximum-a-posteriori (MAP) reconstruction algorithm. In this contribution, we use the PET-SORTEO Monte Carlo simulator to evaluate the quantitative accuracy reached by three different anatomical priors when reconstructing positron emission tomography (PET) brain images, using volumetric magnetic resonance imaging (MRI) to provide the anatomical information. The priors are: 1) a prior especially developed for FDG PET brain imaging, which relies on a segmentation of the MR-image (Baete , 2004); 2) the joint entropy-prior (Nuyts, 2007); 3) a prior that encourages smoothness within a position dependent neighborhood, computed from the MR-image. The latter prior was recently proposed by our group in (Vunckx and Nuyts, 2010), and was based on the prior presented by Bowsher (2004). The two latter priors do not rely on an explicit segmentation, which makes them more generally applicable than a segmentation-based prior. All three priors produced a compromise between noise and bias that was clearly better than that obtained with postsmoothed maximum likelihood expectation maximization (MLEM) or MAP with a relative difference prior. The performance of the joint entropy prior was slightly worse than that of the other two priors. The performance of the segmentation-based prior is quite sensitive to the accuracy of the segmentation. In contrast to the joint entropy-prior, the Bowsher-prior is easily tuned and does not suffer from convergence problems.

The purpose of the work described in this paper was the development of a framework for the creation of a realistic positron emission tomography (PET) simulated database incorporating patient-specific variability. The ground truth used was therefore based on clinical PET/computed tomography (CT) data of oncology patients. In the first step, the NURBS-based cardiac-torso phantom was adapted to the patient's CT acquisitions to reproduce their specific anatomy while the corresponding PET acquisitions were used to derive the activity distribution of each organ of interest. Secondly, realistic tumor shapes with homogeneous or heterogeneous activity distributions were modeled based on segmentation of the PET tumor volume and incorporated in the patient-specific models obtained at the first step. Lastly, patient-specific respiratory motion was also modeled. The derived patient-specific models were subsequently combined with the PET SORTEO Monte Carlo simulation tool for the simulation of the whole-body PET acquisition process. The accuracy of the simulated datasets was assessed in comparison to the original clinical patient images. In addition, a couple of applications for such simulated images were also demonstrated. Future work will focus on the creation of a comprehensive database of simulated raw data and reconstructed whole-body images, facilitating the rigorous evaluation of image-processing algorithms in PET for oncology applications.

This paper describes the construction of simulated SPECT and MRI databases that account for realistic anatomical and functional variability. The data is used as a gold-standard to evaluate four SPECT/MRI similarity-based registration methods. Simulation realism was accounted for using accurate physical models of data generation and acquisition. MRI and SPECT simulations were generated from three subjects to take into account inter-subject anatomical variability. Functional SPECT data were computed from six functional models of brain perfusion. Previous models of normal perfusion and ictal perfusion observed in Mesial Temporal Lobe Epilepsy (MTLE) were considered to generate functional variability. We studied the impact noise and intensity non-uniformity in MRI simulations and SPECT scatter correction may have on registration accuracy. We quantified the amount of registration error caused by anatomical and functional variability. Registration involving ictal data was less accurate than registration involving normal data. MR intensity nonuniformity was the main factor decreasing registration accuracy. The proposed simulated database is promising to evaluate many functional neuroimaging methods, involving MRI and SPECT data.

During the process of validation of a new tracer, estimation of performance and validation of processing algorithms have to be investigated with data sets representative of the ground truth. Because this ground truth is hardly accessible in positron emission tomography (PET), validations of processing algorithms often rely on the use of simulated data sets. Considering that Monte Carlo simulators are very time consuming and are not very easy to use, the building of publicly available databases of simulated PET volumes are becoming highly desirable. We present here the methodology employed for the creation of a database of simulated dynamic [18F]MPPF-PET data, including inter-individual anatomical and biological variability which meets the criteria of a gold standard database as defined by Lehmann: reliance, equivalence, independence, relevance, significance. The assessment of the realism of the built database against actual MPPF PET data is also presented here. Whereas the database was specifically created for the investigations of quantification of activity and binding of ligand-receptor with the [18F]MPPF PET tracer, it may serve the community with countless purposes. The full strength of this database, does not only stem from the knowledge of important information such as the true activity map and underlying anatomical data, but also from the possibility to fully control the biological difference between sets of simulated PET data. Indeed, time activity curves included in the simulated data sets are controlled by a multicompartmental model of ligand-receptor exchanges. This latter feature is of a great interest in the context of the improvement of the detectability of biological variation in PET.