Voxel-based comparison of state-of-the-art reconstruction algorithms for 18F-FDG PET brain imaging using simulated and clinical data

AI for PET image reconstruction

Image reconstruction for positron emission tomography (PET) has been developed over many decades, with advances coming from improved modelling of the data statistics and improved modelling of the imaging physics. However, high noise and limited spatial resolution have remained issues in PET imaging, and state-of-the-art PET reconstruction has started to exploit other medical imaging modalities (such as MRI) to assist in noise reduction and enhancement of PET's spatial resolution. Nonetheless, there is an ongoing drive towards not only improving image quality, but also reducing the injected radiation dose and reducing scanning times. While the arrival of new PET scanners (such as total body PET) is helping, there is always a need to improve reconstructed image quality due to the time and count limited imaging conditions. Artificial intelligence (AI) methods are now at the frontier of research for PET image reconstruction. While AI can learn the imaging physics as well as the noise in the data (when given sufficient examples), one of the most common uses of AI arises from exploiting databases of high-quality reference examples, to provide advanced noise compensation and resolution recovery. There are three main AI reconstruction approaches: (i) direct data-driven AI methods which rely on supervised learning from reference data, (ii) iterative (unrolled) methods which combine our physics and statistical models with AI learning from data, and (iii) methods which exploit AI with our known models, but crucially can offer benefits even in the absence of any example training data whatsoever. This article reviews these methods, considering opportunities and challenges of AI for PET reconstruction.

Feasibility of dose reduction for [18F]FDG-PET/MR imaging of patients with non-lesional epilepsy

The aim of the study was to evaluate the effect of reduced injected [18F]FDG activity levels on the quantitative and diagnostic accuracy of PET images of patients with non-lesional epilepsy (NLE).Nine healthy volunteers and nine patients with NLE underwent 60-min dynamic list-mode (LM) scans on a fully-integrated PET/MRI system. Injected FDG activity levels were reduced virtually by randomly removing counts from the last 10-min of the LM data, so as to simulate the following activity levels: 50 %, 35 %, 20 %, and 10 % of the original activity. Four image reconstructions were evaluated: standard OSEM, OSEM with resolution recovery (PSF), the A-MAP, and the Asymmetrical Bowsher (AsymBowsher) algorithms. For the A-MAP algorithms, two weights were selected (low and high). Image contrast and noise levels were evaluated for all subjects while the lesion-to-background ratio (L/B) was only evaluated for patients. Patient images were scored by a Nuclear Medicine physician on a 5-point scale to assess clinical impression associated with the various reconstruction algorithms.The image contrast and L/B ratio characterizing all four reconstruction algorithms were similar, except for reconstructions based on only 10 % of total counts. Based on clinical impression, images with diagnostic quality can be achieved with as low as 35 % of the standard injected activity. The selection of algorithms utilizing an anatomical prior did not provide a significant advantage for clinical readings, despite a small improvement in L/B (< 5 %) using the A-MAP and AsymBowsher reconstruction algorithms.In patients with NLE who are undergoing [18F]FDG-PET/MR imaging, the injected [18F]FDG activity can be reduced to 35 % of the original dose levels without compromising.

Nuclear imaging for localization and surgical outcome prediction in epilepsy: A review of latest discoveries and future perspectives

Background

Epilepsy is one of the most common neurological disorders. Approximately, one-third of patients with epilepsy have seizures refractory to antiepileptic drugs and further require surgical removal of the epileptogenic region. In the last decade, there have been many recent developments in radiopharmaceuticals, novel image analysis techniques, and new software for an epileptogenic zone (EZ) localization.

Objectives

Recently, we provided the latest discoveries, current challenges, and future perspectives in the field of positron emission tomography (PET) and single-photon emission computed tomography (SPECT) in epilepsy.

Methods

We searched for relevant articles published in MEDLINE and CENTRAL from July 2012 to July 2022. A systematic literature review based on the Preferred Reporting Items for Systematic Reviews and Meta-Analysis was conducted using the keywords "Epilepsy" and "PET or SPECT." We included both prospective and retrospective studies. Studies with preclinical subjects or not focusing on EZ localization or surgical outcome prediction using recently developed PET radiopharmaceuticals, novel image analysis techniques, and new software were excluded from the review. The remaining 162 articles were reviewed.

Results

We first present recent findings and developments in PET radiopharmaceuticals. Second, we present novel image analysis techniques and new software in the last decade for EZ localization. Finally, we summarize the overall findings and discuss future perspectives in the field of PET and SPECT in epilepsy.

Conclusion

Combining new radiopharmaceutical development, new indications, new techniques, and software improves EZ localization and provides a better understanding of epilepsy. These have proven not to only predict prognosis but also to improve the outcome of epilepsy surgery.

Comparison of an in-house acquired brain F-18 FDG PET normal database with commercially available normal data

INTRODUCTION

Current guidelines recommend the use of semiautomated assessment of F-18 FDG PET brain studies. Accuracy is influenced by the normal data, which requires knowledge of the included subjects and how they were acquired. Due to confidentiality, such information is often not completely disclosed. Our aim was to determine the variation in FDG uptake between several commercially available and our in-house normal database.

METHODS

Our database contains 83 healthy subjects. Outlier detection using SPM further ensured normality, resulting in exclusion of three subjects. The remaining 80 subjects were analyzed using three commercially available software packages. Z-score data per patient and per lobe were extracted and pooled in predefined age groups (18-40, 41-60 and 61-80 years old) with a calculation of mean Z-scores and SD. Correlation between Z-score output of different software was investigated.

RESULTS

In the 18-40 years age group, frontotemporal hypermetabolism was found with all software. Decreased cerebellar uptake was found with two software packages. Mean Z-scores are closer to zero in the 41-60 years age group compared to the younger group, and mostly within the normal range in the 61-80 years age group with all software. A moderate to high linear correlation between Z-score output was found, but individual Z-scores varied widely.

CONCLUSIONS

The three software packages yielded varying Z-score output, partially explained by an age mismatch between our subjects and subjects in their normal databases. A definitive explanation for the remaining differences is lacking. This emphasizes the importance of age-matched normal data and knowledge of the included databases to allow adequate preprocessing.

Quantitative Rodent Brain Receptor Imaging

Positron emission tomography (PET) is a non-invasive imaging technology employed to describe metabolic, physiological, and biochemical processes in vivo. These include receptor availability, metabolic changes, neurotransmitter release, and alterations of gene expression in the brain. Since the introduction of dedicated small-animal PET systems along with the development of many novel PET imaging probes, the number of PET studies using rats and mice in basic biomedical research tremendously increased over the last decade. This article reviews challenges and advances of quantitative rodent brain imaging to make the readers aware of its physical limitations, as well as to inspire them for its potential applications in preclinical research. In the first section, we briefly discuss the limitations of small-animal PET systems in terms of spatial resolution and sensitivity and point to possible improvements in detector development. In addition, different acquisition and post-processing methods used in rodent PET studies are summarized. We further discuss factors influencing the test-retest variability in small-animal PET studies, e.g., different receptor quantification methodologies which have been mainly translated from human to rodent receptor studies to determine the binding potential and changes of receptor availability and radioligand affinity. We further review different kinetic modeling approaches to obtain quantitative binding data in rodents and PET studies focusing on the quantification of endogenous neurotransmitter release using pharmacological interventions. While several studies have focused on the dopamine system due to the availability of several PET tracers which are sensitive to dopamine release, other neurotransmitter systems have become more and more into focus and are described in this review, as well. We further provide an overview of latest genome engineering technologies, including the CRISPR/Cas9 and DREADD systems that may advance our understanding of brain disorders and function and how imaging has been successfully applied to animal models of human brain disorders. Finally, we review the strengths and opportunities of simultaneous PET/magnetic resonance imaging systems to study drug-receptor interactions and challenges for the translation of PET results from bench to bedside.

Approximating anatomically-guided PET reconstruction in image space using a convolutional neural network

In the last two decades, it has been shown that anatomically-guided PET reconstruction can lead to improved bias-noise characteristics in brain PET imaging. However, despite promising results in simulations and first studies, anatomically-guided PET reconstructions are not yet available for use in routine clinical because of several reasons. In light of this, we investigate whether the improvements of anatomically-guided PET reconstruction methods can be achieved entirely in the image domain with a convolutional neural network (CNN). An entirely image-based CNN post-reconstruction approach has the advantage that no access to PET raw data is needed and, moreover, that the prediction times of trained CNNs are extremely fast on state of the art GPUs which will substantially facilitate the evaluation, fine-tuning and application of anatomically-guided PET reconstruction in real-world clinical settings. In this work, we demonstrate that anatomically-guided PET reconstruction using the asymmetric Bowsher prior can be well-approximated by a purely shift-invariant convolutional neural network in image space allowing the generation of anatomically-guided PET images in almost real-time. We show that by applying dedicated data augmentation techniques in the training phase, in which 16 [18F]FDG and 10 [18F]PE2I data sets were used, lead to a CNN that is robust against the used PET tracer, the noise level of the input PET images and the input MRI contrast. A detailed analysis of our CNN in 36 [18F]FDG, 18 [18F]PE2I, and 7 [18F]FET test data sets demonstrates that the image quality of our trained CNN is very close to the one of the target reconstructions in terms of regional mean recovery and regional structural similarity.

3D Auto-Context-Based Locality Adaptive Multi-Modality GANs for PET Synthesis

Positron emission tomography (PET) has been substantially used recently. To minimize the potential health risk caused by the tracer radiation inherent to PET scans, it is of great interest to synthesize the high-quality PET image from the low-dose one to reduce the radiation exposure. In this paper, we propose a 3D auto-context-based locality adaptive multi-modality generative adversarial networks model (LA-GANs) to synthesize the high-quality FDG PET image from the low-dose one with the accompanying MRI images that provide anatomical information. Our work has four contributions. First, different from the traditional methods that treat each image modality as an input channel and apply the same kernel to convolve the whole image, we argue that the contributions of different modalities could vary at different image locations, and therefore a unified kernel for a whole image is not optimal. To address this issue, we propose a locality adaptive strategy for multi-modality fusion. Second, we utilize 1 ×1 ×1 kernel to learn this locality adaptive fusion so that the number of additional parameters incurred by our method is kept minimum. Third, the proposed locality adaptive fusion mechanism is learned jointly with the PET image synthesis in a 3D conditional GANs model, which generates high-quality PET images by employing large-sized image patches and hierarchical features. Fourth, we apply the auto-context strategy to our scheme and propose an auto-context LA-GANs model to further refine the quality of synthesized images. Experimental results show that our method outperforms the traditional multi-modality fusion methods used in deep networks, as well as the state-of-the-art PET estimation approaches.

Correlation of neuropsychological and metabolic changes after epilepsy surgery in patients with left mesial temporal lobe epilepsy with hippocampal sclerosis

Background

Epilepsy surgery often causes changes in cognition and cerebral glucose metabolism. Our aim was to explore relationships between pre- and postoperative cerebral metabolism as measured with ¹⁸F-fluorodeoxyglucose positron emission tomography (FDG-PET) and neuropsychological test scores in patients with left mesial temporal lobe epilepsy with hippocampal sclerosis (MTLE-HS), who were rendered seizure-free after epilepsy surgery.

Results

Thirteen patients were included. All had neuropsychological testing and an interictal FDG-PET scan of the brain pre- and postoperative. Correlations between changes in neuropsychological test scores and metabolism were examined using statistical parametric mapping (SPM). There were no significant changes in the neuropsychological test scores pre- and postoperatively at the group level. Decreased metabolism was observed in the left mesial temporal regions and occipital lobe. Increased metabolism was observed in the bi-frontal and right parietal lobes, temporal lobes, occipital lobes, thalamus, cerebellum, and vermis. In these regions, we did not find a correlation between changes in metabolism and neuropsychological test scores. A significant negative correlation, however, was found between metabolic changes in the precuneus and Boston Naming Test (BNT) scores.

Conclusions

There are significant metabolic decreases in the left mesial temporal regions and increases in the bi-frontal lobes; right parietal, temporal, and occipital lobes; right thalamus; cerebellum; and vermis in patients with left MTLE-HS who were rendered seizure-free after epilepsy surgery. We could not confirm that these changes translate into significant cognitive changes. A significant negative correlation was found between changes in confrontation naming and changes in metabolism in the precuneus. We speculate that the precuneus may play a compensatory role in patients with postoperative naming difficulties after left TLE surgery. Understanding of these neural mechanisms may aid in designing cognitive rehabilitation strategies.

Electronic supplementary material

The online version of this article (10.1186/s13550-018-0385-5) contains supplementary material, which is available to authorized users.

Evaluation of Parallel Level Sets and Bowsher's Method as Segmentation-Free Anatomical Priors for Time-of-Flight PET Reconstruction

In this article, we evaluate Parallel Level Sets (PLS) and Bowsher's method as segmentation-free anatomical priors for regularized brain positron emission tomography (PET) reconstruction. We derive the proximity operators for two PLS priors and use the EM-TV algorithm in combination with the first order primal-dual algorithm by Chambolle and Pock to solve the non-smooth optimization problem for PET reconstruction with PLS regularization. In addition, we compare the performance of two PLS versions against the symmetric and asymmetric Bowsher priors with quadratic and relative difference penalty function. For this aim, we first evaluate reconstructions of 30 noise realizations of simulated PET data derived from a real hybrid positron emission tomography/magnetic resonance imaging (PET/MR) acquisition in terms of regional bias and noise. Second, we evaluate reconstructions of a real brain PET/MR data set acquired on a GE Signa time-of-flight PET/MR in a similar way. The reconstructions of simulated and real 3D PET/MR data show that all priors were superior to post-smoothed maximum likelihood expectation maximization with ordered subsets (OSEM) in terms of bias-noise characteristics in different regions of interest where the PET uptake follows anatomical boundaries. Our implementation of the asymmetric Bowsher prior showed slightly superior performance compared with the two versions of PLS and the symmetric Bowsher prior. At very high regularization weights, all investigated anatomical priors suffer from the transfer of non-shared gradients.

Advances in PET/MR instrumentation and image reconstruction

The combination of positron emission tomography (PET) and MRI has attracted the attention of researchers in the past approximately 20 years in small-animal imaging and more recently in clinical research. The combination of PET/MRI allows researchers to explore clinical and research questions in a wide number of fields, some of which are briefly mentioned here. An important number of groups have developed different concepts to tackle the problems that PET instrumentation poses to the exposition of electromagnetic fields. We have described most of these research developments in preclinical and clinical experiments, including the few commercial scanners available. From the software perspective, an important number of algorithms have been developed to address the attenuation correction issue and to exploit the possibility that MRI provides for motion correction and quantitative image reconstruction, especially parametric modelling of radiopharmaceutical kinetics. In this work, we give an overview of some exemplar applications of simultaneous PET/MRI, together with technological hardware and software developments.

Lesion quantification and detection in myocardial (18)F-FDG PET using edge-preserving priors and anatomical information from CT and MRI: a simulation study

Background

The limited spatial resolution of the clinical PET scanners results in image blurring and does not allow for accurate quantification of very thin or small structures (known as partial volume effect). In cardiac imaging, clinically relevant questions, e.g. to accurately define the extent or the residual metabolic activity of scarred myocardial tissue, could benefit from partial volume correction (PVC) techniques.

The use of high-resolution anatomical information for improved reconstruction of the PET datasets has been successfully applied in other anatomical regions. However, several concerns linked to the use of any kind of anatomical information for PVC on cardiac datasets arise. The moving nature of the heart, coupled with the possibly non-simultaneous acquisition of the anatomical and the activity datasets, is likely to introduce discrepancies between the PET and the anatomical image, that in turn might mislead lesion quantification and detection. Non-anatomical (edge-preserving) priors could represent a viable alternative for PVC in this case.

In this work, we investigate and compare the regularizing effect of different anatomical and non-anatomical priors applied during maximum-a-posteriori (MAP) reconstruction of cardiac PET datasets. The focus of this paper is on accurate quantification and lesion detection in myocardial ¹⁸F-FDG PET.

Methods

Simulated datasets, obtained with the XCAT software, are reconstructed with different algorithms and are quantitatively analysed.

Results

The results of this simulation study show a superiority of the anatomical prior when an ideal, perfectly matching anatomy is used. The anatomical information must clearly differentiate between normal and scarred myocardial tissue for the PVC to be successful. In case of mismatched or missing anatomical information, the quality of the anatomy-based MAP reconstructions decreases, affecting both overall image quality and lesion quantification. The edge-preserving priors produce reconstructions with good noise properties and recovery of activity, with the advantage of not relying on an external, additional scan for anatomy.

Conclusions

The performance of edge-preserving priors is acceptable but inferior to those of a well-applied anatomical prior that differentiates between lesion and normal tissue, in the detection and quantification of a lesion in the reconstructed images. When considering bull’s eye plots, all of the tested MAP algorithms produced comparable results.

Fast Patch-Based Pseudo-CT Synthesis from T1-Weighted MR Images for PET/MR Attenuation Correction in Brain Studies

Attenuation correction in hybrid PET/MR scanners is still a challenging task. This paper describes a methodology for synthesizing a pseudo-CT volume from a single T1-weighted volume, thus allowing us to create accurate attenuation correction maps.

METHODS

We propose a fast pseudo-CT volume generation from a patient-specific MR T1-weighted image using a groupwise patch-based approach and an MRI-CT atlas dictionary. For every voxel in the input MR image, we compute the similarity of the patch containing that voxel to the patches of all MR images in the database that lie in a certain anatomic neighborhood. The pseudo-CT volume is obtained as a local weighted linear combination of the CT values of the corresponding patches. The algorithm was implemented in a graphical processing unit (GPU).

RESULTS

We evaluated our method both qualitatively and quantitatively for PET/MR correction. The approach performed successfully in all cases considered. We compared the SUVs of the PET image obtained after attenuation correction using the patient-specific CT volume and using the corresponding computed pseudo-CT volume. The patient-specific correlation between SUV obtained with both methods was high (R(2) = 0.9980, P < 0.0001), and the Bland-Altman test showed that the average of the differences was low (0.0006 ± 0.0594). A region-of-interest analysis was also performed. The correlation between SUVmean and SUVmax for every region was high (R(2) = 0.9989, P < 0.0001, and R(2) = 0.9904, P < 0.0001, respectively).

CONCLUSION

The results indicate that our method can accurately approximate the patient-specific CT volume and serves as a potential solution for accurate attenuation correction in hybrid PET/MR systems. The quality of the corrected PET scan using our pseudo-CT volume is comparable to having acquired a patient-specific CT scan, thus improving the results obtained with the ultrashort-echo-time-based attenuation correction maps currently used in the scanner. The GPU implementation substantially decreases computational time, making the approach suitable for real applications.