PET image reconstruction using kernel method

The Evolution of Image Reconstruction in PET: From Filtered Back-Projection to Artificial Intelligence

PET can provide functional images revealing physiologic processes in vivo. Although PET has many applications, there are still some limitations that compromise its precision: the absorption of photons in the body causes signal attenuation; the dead-time limit of system components leads to the loss of the count rate; the scattered and random events received by the detector introduce additional noise; the characteristics of the detector limit the spatial resolution; and the low signal-to-noise ratio caused by the scan-time limit (eg, dynamic scans) and dose concern. The early PET reconstruction methods are analytical approaches based on an idealized mathematical model.

Anatomically aided PET image reconstruction using deep neural networks

PURPOSE

The developments of PET/CT and PET/MR scanners provide opportunities for improving PET image quality by using anatomical information. In this paper, we propose a novel co-learning three-dimensional (3D) convolutional neural network (CNN) to extract modality-specific features from PET/CT image pairs and integrate complementary features into an iterative reconstruction framework to improve PET image reconstruction.

METHODS

We used a pretrained deep neural network to represent PET images. The network was trained using low-count PET and CT image pairs as inputs and high-count PET images as labels. This network was then incorporated into a constrained maximum likelihood framework to regularize PET image reconstruction. Two different network structures were investigated for the integration of anatomical information from CT images. One was a multichannel CNN, which treated PET and CT volumes as separate channels of the input. The other one was multibranch CNN, which implemented separate encoders for PET and CT images to extract latent features and fed the combined latent features into a decoder. Using computer-based Monte Carlo simulations and two real patient datasets, the proposed method has been compared with existing methods, including the maximum likelihood expectation maximization (MLEM) reconstruction, a kernel-based reconstruction and a CNN-based deep penalty method with and without anatomical guidance.

RESULTS

Reconstructed images showed that the proposed constrained ML reconstruction approach produced higher quality images than the competing methods. The tumors in the lung region have higher contrast in the proposed constrained ML reconstruction than in the CNN-based deep penalty reconstruction. The image quality was further improved by incorporating the anatomical information. Moreover, the liver standard deviation was lower in the proposed approach than all the competing methods at a matched lesion contrast.

CONCLUSIONS

The supervised co-learning strategy can improve the performance of constrained maximum likelihood reconstruction. Compared with existing techniques, the proposed method produced a better lesion contrast versus background standard deviation trade-off curve, which can potentially improve lesion detection.

Modified kernel MLAA using autoencoder for PET-enabled dual-energy CT

Combined use of PET and dual-energy CT provides complementary information for multi-parametric imaging. PET-enabled dual-energy CT combines a low-energy X-ray CT image with a high-energy γ-ray CT (GCT) image reconstructed from time-of-flight PET emission data to enable dual-energy CT material decomposition on a PET/CT scanner. The maximum-likelihood attenuation and activity (MLAA) algorithm has been used for GCT reconstruction but suffers from noise. Kernel MLAA exploits an X-ray CT image prior through the kernel framework to guide GCT reconstruction and has demonstrated substantial improvements in noise suppression. However, similar to other kernel methods for image reconstruction, the existing kernel MLAA uses image intensity-based features to construct the kernel representation, which is not always robust and may lead to suboptimal reconstruction with artefacts. In this paper, we propose a modified kernel method by using an autoencoder convolutional neural network (CNN) to extract an intrinsic feature set from the X-ray CT image prior. A computer simulation study was conducted to compare the autoencoder CNN-derived feature representation with raw image patches for evaluation of kernel MLAA for GCT image reconstruction and dual-energy multi-material decomposition. The results show that the autoencoder kernel MLAA method can achieve a significant image quality improvement for GCT and material decomposition as compared to the existing kernel MLAA algorithm. A weakness of the proposed method is its potential over-smoothness in a bone region, indicating the importance of further optimization in future work. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 2'.

Multiparametric cardiac18F-FDG PET in humans: pilot comparison of FDG delivery rate with82Rb myocardial blood flow

Myocardial blood flow (MBF) and flow reserve are usually quantified in the clinic with positron emission tomography (PET) using a perfusion-specific radiotracer (e.g.⁸²Rb-chloride). However, the clinical accessibility of existing perfusion tracers remains limited. Meanwhile,¹⁸F-fluorodeoxyglucose (FDG) is a commonly used radiotracer for PET metabolic imaging without similar limitations. In this paper, we explore the potential of¹⁸F-FDG for myocardial perfusion imaging by comparing the myocardial FDG delivery rateK₁with MBF as determined by dynamic⁸²Rb PET in fourteen human subjects with heart disease. Two sets of FDGK₁were derived from one-hour dynamic FDG scans. One was the original FDGK₁estimates and the other was the correspondingK₁values that were linearly normalized for blood glucose levels. A generalized Renkin-Crone model was used to fit FDGK₁with Rb MBF, which then allowed for a nonlinear extraction fraction correction for converting FDGK₁to MBF. The linear correlation between FDG-derived MBF and Rb MBF was moderate (r= 0.79) before the glucose normalization and became much improved (r> 0.9) after glucose normalization. The extraction fraction of FDG was also similar to that of Rb-chloride in the myocardium. The results from this pilot study suggest that dynamic cardiac FDG-PET with tracer kinetic modeling has the potential to provide MBF in addition to its conventional use for metabolic imaging.

MRI-aided kernel PET image reconstruction method based on texture features

We investigate the reconstruction of low-count positron emission tomography (PET) projection, which is an important, but challenging, task. Using the texture feature extraction method of radiomics, i.e. the gray-level co-occurrence matrix (GLCM), texture features can be extracted from magnetic resonance imaging images with high-spatial resolution. In this work, we propose a kernel reconstruction method combining autocorrelation texture features derived from the GLCM. The new kernel function includes the correlations of both the intensity and texture features from the prior image. By regarding the GLCM as a discrete approximation of a probability density function, the asymptotically gray-level-invariant autocorrelation texture feature is generated, which can maintain the accuracy of texture features extracted from small image regions by reducing the number of quantized image gray levels. A computer simulation shows that the proposed method can effectively reduce the noise in the reconstructed image compared to the maximum likelihood expectation maximum method and improve the image quality and tumor region accuracy compared to the original kernel method for low-count PET reconstruction. A simulation study on clinical patient images also shows that the proposed method can improve the whole image quality and that the reconstruction of a high-uptake lesion is more accurate than that achieved by the original kernel method.

Technical note: A preliminary study of dual-tracer PET image reconstruction guided by FDG and/or MR kernels

PURPOSE

Clinically, single radiotracer positron emission tomography (PET) imaging is a commonly used examination method; however, since each radioactive tracer reflects the information of only one kind of cell, it easily causes false negatives or false positives in disease diagnosis. Therefore, reasonably combining two or more radiotracers is recommended to improve the accuracy of diagnosis and the sensitivity and specificity of the disease when conditions permit.

METHODS

This paper proposes incorporating ¹⁸ F-fluorodeoxyglucose (FDG) as a higher-quality PET image to guide the reconstruction of other lower-count ¹¹ C-methionine (MET) PET datasets to compensate for the lower image quality by a popular kernel algorithm. Specifically, the FDG prior is needed to extract kernel features, and these features were used to build a kernel matrix using a k-nearest-neighbor (kNN) search for MET image reconstruction. We created a 2-D brain phantom to validate the proposed method by simulating sinogram data containing Poisson random noise and quantitatively compared the performance of the proposed FDG-guided kernelized expectation maximization (KEM) method with the performance of Gaussian and non-local means (NLM) smoothed maximum likelihood expectation maximization (MLEM), MR-guided KEM, and multi-guided-S KEM algorithms. Mismatch experiments between FDG/MR and MET data were also carried out to investigate the outcomes of possible clinical situations.

RESULTS

In the simulation study, the proposed method outperformed the other algorithms by at least 3.11% in the signal-to-noise ratio (SNR) and 0.68% in the contrast recovery coefficient (CRC), and it reduced the mean absolute error (MAE) by 8.07%. Regarding the tumor in the reconstructed image, the proposed method contained more pathological information. Furthermore, the proposed method was still superior to the MR-guided KEM method in the mismatch experiments.

CONCLUSIONS

The proposed FDG-guided KEM algorithm can effectively utilize and compensate for the tissue metabolism information obtained from dual-tracer PET to maximize the advantages of PET imaging.

Rapid high-quality PET Patlak parametric image generation based on direct reconstruction and temporal nonlocal neural network

Parametric imaging based on dynamic positron emission tomography (PET) has wide applications in neurology. Compared to indirect methods, direct reconstruction methods, which reconstruct parametric images directly from the raw PET data, have superior image quality due to better noise modeling and richer information extracted from the PET raw data. For low-dose scenarios, the advantages of direct methods are more obvious. However, the wide adoption of direct reconstruction is inevitably impeded by the excessive computational demand and deficiency of the accessible raw data. In addition, motion modeling inside dynamic PET image reconstruction raises more computational challenges for direct reconstruction methods. In this work, we focused on the ¹⁸F-FDG Patlak model, and proposed a data-driven approach which can estimate the motion corrected full-dose direct Patlak images from the dynamic PET reconstruction series, based on a proposed novel temporal non-local convolutional neural network. During network training, direct reconstruction with motion correction based on full-dose dynamic PET sinograms was performed to obtain the training labels. The reconstructed full-dose /low-dose dynamic PET images were supplied as the network input. In addition, a temporal non-local block based on the dynamic PET images was proposed to better recover the structural information and reduce the image noise. During testing, the proposed network can directly output high-quality Patlak parametric images from the full-dose /low-dose dynamic PET images in seconds. Experiments based on 15 full-dose and 15 low-dose ¹⁸F-FDG brain datasets were conducted and analyzed to validate the feasibility of the proposed framework. Results show that the proposed framework can generate better image quality than reference methods.

Synergistic tomographic image reconstruction: part 1

This special issue focuses on synergistic tomographic image reconstruction in a range of contributions in multiple disciplines and various application areas. The topic of image reconstruction covers substantial inverse problems (Mathematics) which are tackled with various methods including statistical approaches (e.g. Bayesian methods, Monte Carlo) and computational approaches (e.g. machine learning, computational modelling, simulations). The issue is separated in two volumes. This volume focuses mainly on algorithms and methods. Some of the articles will demonstrate their utility on real-world challenges, either medical applications (e.g. cardiovascular diseases, proton therapy planning) or applications in material sciences (e.g. material decomposition and characterization). One of the desired outcomes of the special issue is to bring together different scientific communities which do not usually interact as they do not share the same platforms (such as journals and conferences). This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 1'.

Improved identification of abdominal aortic aneurysm using the Kernelized Expectation Maximization algorithm

Abdominal aortic aneurysm (AAA) monitoring and risk of rupture is currently assumed to be correlated with the aneurysm diameter. Aneurysm growth, however, has been demonstrated to be unpredictable. Using PET to measure uptake of [¹⁸F]-NaF in calcified lesions of the abdominal aorta has been shown to be useful for identifying AAA and to predict its growth. The PET low spatial resolution, however, can affect the accuracy of the diagnosis. Advanced edge-preserving reconstruction algorithms can overcome this issue. The kernel method has been demonstrated to provide noise suppression while retaining emission and edge information. Nevertheless, these findings were obtained using simulations, phantoms and a limited amount of patient data. In this study, the authors aim to investigate the usefulness of the anatomically guided kernelized expectation maximization (KEM) and the hybrid KEM (HKEM) methods and to judge the statistical significance of the related improvements. Sixty-one datasets of patients with AAA and 11 from control patients were reconstructed with ordered subsets expectation maximization (OSEM), HKEM and KEM and the analysis was carried out using the target-to-blood-pool ratio, and a series of statistical tests. The results show that all algorithms have similar diagnostic power, but HKEM and KEM can significantly recover uptake of lesions and improve the accuracy of the diagnosis by up to 22% compared to OSEM. The same improvements are likely to be obtained in clinical applications based on the quantification of small lesions, like for example cancer. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 1'.

(An overview of) Synergistic reconstruction for multimodality/multichannel imaging methods

Imaging is omnipresent in modern society with imaging devices based on a zoo of physical principles, probing a specimen across different wavelengths, energies and time. Recent years have seen a change in the imaging landscape with more and more imaging devices combining that which previously was used separately. Motivated by these hardware developments, an ever increasing set of mathematical ideas is appearing regarding how data from different imaging modalities or channels can be synergistically combined in the image reconstruction process, exploiting structural and/or functional correlations between the multiple images. Here we review these developments, give pointers to important challenges and provide an outlook as to how the field may develop in the forthcoming years. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 1'.

Direct and indirect parameter imaging methods for dynamic PET

The method of reconstructing parametric images from dynamic positron emission tomography (PET) data with the linear Patlak model has been widely used in scientific research and clinical practice. Whether for direct or indirect image reconstruction, researchers have deeply investigated the associated methods and effects. Among the existing methods, the traditional maximum likelihood expectation maximization (MLEM) reconstruction algorithm is fast but produces a substantial amount of noise. If the parameter images obtained by the MLEM algorithm are postfiltered, a large amount of image edge information is lost. Additionally, although the kernel method has a better noise reduction effect, its calculation costs are very high due to the complexity of the algorithm. Therefore, to obtain parametric images with a high signal-to-noise ratio (SNR) and good retention of detailed information, here, we use guided kernel means (GKM) and dynamic PET image information to conduct guided filtering and perform parametric image reconstruction. We apply this method to direct and indirect reconstruction, and through computer simulations, we show that our proposed method has higher identifiability and a greater SNR than conventional direct and indirect reconstruction methods. We also show that our method produces better images with direct than with indirect reconstruction.

Applications of artificial intelligence in nuclear medicine image generation

Recently, the application of artificial intelligence (AI) in medical imaging (including nuclear medicine imaging) has rapidly developed. Most AI applications in nuclear medicine imaging have focused on the diagnosis, treatment monitoring, and correlation analyses with pathology or specific gene mutation. It can also be used for image generation to shorten the time of image acquisition, reduce the dose of injected tracer, and enhance image quality. This work provides an overview of the application of AI in image generation for single-photon emission computed tomography (SPECT) and positron emission tomography (PET) either without or with anatomical information [CT or magnetic resonance imaging (MRI)]. This review focused on four aspects, including imaging physics, image reconstruction, image postprocessing, and internal dosimetry. AI application in generating attenuation map, estimating scatter events, boosting image quality, and predicting internal dose map is summarized and discussed.

Low activity [11C]raclopride kinetic modeling in the mouse brain using the spatiotemporal kernel method

Depending on the molar activity of the tracer, the maximal allowable injected activity in mouse brain PET studies can be extremely low in order to avoid receptor saturation. Therefore, a high level of noise can be present in the image. We investigate several dynamic PET reconstruction methods in reduced counts, or equivalently in reduced injected activity, data exemplified in [¹¹C]racloprideBP_NDandR1quantification using the simplified reference tissue model (SRTM). We compared independent frame reconstruction (IFR), post-reconstruction HYPR denoising (IFR + HYPR), direct reconstruction using the SRTM model (DIR-SRTM), and the spatial (KERS) and spatiotemporal kernel reconstruction (KERST). Additionally, HYPR denoising of the frames used as features for the calculation of the spatial kernel matrix, was investigated (KERS-HYPR and KERST-HYPR).In vivodata of 11 mice, was used to generate list-mode data for five reduced count levels corresponding to reductions by a factor 4, 8, 12, 16 and 32 (equivalently 2.07, 1.04, 0.691, 0.518, and 0.260 MBq). Correlation of regionalBP_NDandR1values (reduced versus full counts reconstructions) was high (r > 0.94) for all methods, with KERS-HYPR and KERST-HYPR reaching the highest correlation (r > 0.96). Among methods with regularization, DIR-SRTM showed the largest variability inBP_ND(Bland-Altman SD from 3.0% to 12%), while IFR showed it forR1(5.1%-14.6%). KERST and KERST-HYPR were the only methods with Bland-Altman bias and SD below 5% for noise level up to a reduction factor of 16. At the voxel level,BP_NDandR1correlation was gradually decreased with increasing noise, with the largest correlation (BP_NDr > 0.88,R1r > 0.62) for KERS-HYPR and KERST-HYPR. The spatial and the spatiotemporal kernel methods performed similarly, while using only temporal regularization with direct reconstruction showed more variability. AlthoughR1 values present noise, using the spatiotemporal kernel reconstruction, accurate estimates of binding potential could be obtained with mouse injected activities as low as 0.26-0.518 MBq. This is desirable in order to maintain the tracer kinetics principle in mouse studies.

Dynamic PET image reconstruction utilizing intrinsic data-driven HYPR4D denoising kernel

PURPOSE

Reconstructed PET images are typically noisy, especially in dynamic imaging where the acquired data are divided into several short temporal frames. High noise in the reconstructed images translates to poor precision/reproducibility of image features. One important role of "denoising" is therefore to improve the precision of image features. However, typical denoising methods achieve noise reduction at the expense of accuracy. In this work, we present a novel four-dimensional (4D) denoised image reconstruction framework, which we validate using 4D simulations, experimental phantom, and clinical patient data, to achieve 4D noise reduction while preserving spatiotemporal patterns/minimizing error introduced by denoising.

METHODS

Our proposed 4D denoising operator/kernel is based on HighlY constrained backPRojection (HYPR), which is applied either after each update of OSEM reconstruction of dynamic 4D PET data or within the recently proposed kernelized reconstruction framework inspired by kernel methods in machine learning. Our HYPR4D kernel makes use of the spatiotemporal high frequency features extracted from a 4D composite, generated within the reconstruction, to preserve the spatiotemporal patterns and constrain the 4D noise increment of the image estimate.

RESULTS

Results from simulations, experimental phantom, and patient data showed that the HYPR4D kernel with our proposed 4D composite outperformed other denoising methods, such as the standard OSEM with spatial filter, OSEM with 4D filter, and HYPR kernel method with the conventional 3D composite in conjunction with recently proposed High Temporal Resolution kernel (HYPRC3D-HTR), in terms of 4D noise reduction while preserving the spatiotemporal patterns or 4D resolution within the 4D image estimate. Consequently, the error in outcome measures obtained from the HYPR4D method was less dependent on the region size, contrast, and uniformity/functional patterns within the target structures compared to the other methods. For outcome measures that depend on spatiotemporal tracer uptake patterns such as the nondisplaceable Binding Potential (BP_ND ), the root mean squared error in regional mean of voxel BP_ND values was reduced from ~8% (OSEM with spatial or 4D filter) to ~3% using HYPRC3D-HTR and was further reduced to ~2% using our proposed HYPR4D method for relatively small target structures (~10 mm in diameter). At the voxel level, HYPR4D produced two to four times lower mean absolute error in BP_ND relative to HYPRC3D-HTR.

CONCLUSION

As compared to conventional methods, our proposed HYPR4D method can produce more robust and accurate image features without requiring any prior information.

Quantitative PET in the 2020s: a roadmap

Positron emission tomography (PET) plays an increasingly important role in research and clinical applications, catalysed by remarkable technical advances and a growing appreciation of the need for reliable, sensitive biomarkers of human function in health and disease. Over the last 30 years, a large amount of the physics and engineering effort in PET has been motivated by the dominant clinical application during that period, oncology. This has led to important developments such as PET/CT, whole-body PET, 3D PET, accelerated statistical image reconstruction, and time-of-flight PET. Despite impressive improvements in image quality as a result of these advances, the emphasis on static, semi-quantitative 'hot spot' imaging for oncologic applications has meant that the capability of PET to quantify biologically relevant parameters based on tracer kinetics has not been fully exploited. More recent advances, such as PET/MR and total-body PET, have opened up the ability to address a vast range of new research questions, from which a future expansion of applications and radiotracers appears highly likely. Many of these new applications and tracers will, at least initially, require quantitative analyses that more fully exploit the exquisite sensitivity of PET and the tracer principle on which it is based. It is also expected that they will require more sophisticated quantitative analysis methods than those that are currently available. At the same time, artificial intelligence is revolutionizing data analysis and impacting the relationship between the statistical quality of the acquired data and the information we can extract from the data. In this roadmap, leaders of the key sub-disciplines of the field identify the challenges and opportunities to be addressed over the next ten years that will enable PET to realise its full quantitative potential, initially in research laboratories and, ultimately, in clinical practice.

Micro-Networks for Robust MR-Guided Low Count PET Imaging

Noise suppression is particularly important in low count positron emission tomography (PET) imaging. Post-smoothing (PS) and regularization methods which aim to reduce noise also tend to reduce resolution and introduce bias. Alternatively, anatomical information from another modality such as magnetic resonance (MR) imaging can be used to improve image quality. Convolutional neural networks (CNNs) are particularly well suited to such joint image processing, but usually require large amounts of training data and have mostly been applied outside the field of medical imaging or focus on classification and segmentation, leaving PET image quality improvement relatively understudied. This article proposes the use of a relatively low-complexity CNN (micro-net) as a post-reconstruction MR-guided image processing step to reduce noise and reconstruction artefacts while also improving resolution in low count PET scans. The CNN is designed to be fully 3-D, robust to very limited amounts of training data, and to accept multiple inputs (including competitive denoising methods). Application of the proposed CNN on simulated low (30 M) count data (trained to produce standard (300 M) count reconstructions) results in a 36% lower normalized root mean squared error (NRMSE, calculated over ten realizations against the ground truth) compared to maximum-likelihood expectation maximization (MLEM) used in clinical practice. In contrast, a decrease of only 25% in NRMSE is obtained when an optimized (using knowledge of the ground truth) PS is performed. A 26% NRMSE decrease is obtained with both RM and optimized PS. Similar improvement is also observed for low count real patient datasets. Overfitting to training data is demonstrated to occur as the network size is increased. In an extreme case, a U-net (which produces better predictions for training data) is shown to completely fail on test data due to overfitting to this case of very limited training data. Meanwhile, the resultant images from the proposed CNN (which has low training data requirements) have lower noise, reduced ringing, and partial volume effects, as well as sharper edges and improved resolution compared to conventional MLEM.

Theranostic SPECT reconstruction for improved resolution: application to radionuclide therapy dosimetry

Background

SPECT-derived dose estimates in tissues of diameter less than 3× system resolution are subject to significant losses due to the limited spatial resolution of the gamma camera. Incorporating resolution modelling (RM) into the SPECT reconstruction has been proposed as a possible solution; however, the images produced are prone to noise amplification and Gibbs artefacts. We propose a novel approach to SPECT reconstruction in a theranostic setting, which we term SPECTRE (single photon emission computed theranostic reconstruction); using a diagnostic PET image, with its superior resolution, to guide the SPECT reconstruction of the therapeutic equivalent. This report demonstrates a proof in principle of this approach.

Methods

We have employed the hybrid kernelised expectation maximisation (HKEM) algorithm implemented in STIR, with the aim of producing SPECT images with PET-equivalent resolution. We demonstrate its application in both a dual ⁶⁸Ga/¹⁷⁷Lu IEC phantom study and a clinical example using ⁶⁴Cu/⁶⁷Cu.

Results

SPECTRE is shown to produce images comparable in accuracy and recovery to PET with minimal introduction of artefacts and amplification of noise.

Conclusion

The SPECTRE approach to image reconstruction shows improved quantitative accuracy with a reduction in noise amplification. SPECTRE shows great promise as a method of improving SPECT radioactivity concentrations, directly leading to more accurate dosimetry estimates in small structures and target lesions. Further investigation and optimisation of the algorithm parameters is needed before this reconstruction method can be utilised in a clinical setting.

An improved patch-based regularization method for PET image reconstruction

Background

Statistical reconstruction methods based on penalized maximum likelihood (PML) are being increasingly used in positron emission tomography (PET) imaging to reduce noise and improve image quality. Wang and Qi proposed a patch-based edge-preserving penalties algorithm that can be implemented in three simple steps: a maximum-likelihood expectation-maximization (MLEM) image update, an image smoothing step, and a pixel-by-pixel image fusion step. The pixel-by-pixel image fusion step, which fuses the MLEM updated image and the smoothed image, involves a trade-off between preserving the fine structural features of an image and suppressing noise. Particularly when reconstructing images from low-count data, this step cannot preserve fine structural features in detail. To better preserve these features and accelerate the algorithm convergence, we proposed to improve the patch-based regularization reconstruction method.

Methods

Our improved method involved adding a total variation (TV) regularization step following the MLEM image update in the patch-based algorithm. A feature refinement (FR) step was then used to extract the lost fine structural features from the residual image between the TV regularized image and the fused image based on patch regularization. These structural features would then be added back to the fused image. With the addition of these steps, each iteration of the image should gain more structural information. A brain phantom simulation experiment and a mouse study were conducted to evaluate our proposed improved method. Brain phantom simulation with added noise were used to determine the feasibility of the proposed algorithm and its acceleration of convergence. Data obtained from the mouse study were divided into event count sets to validate the performance of the proposed algorithm when reconstructing images from low-count data. Five criteria were used for quantitative evaluation: signal-to-noise ratio (SNR), covariance (COV), contrast recovery coefficient (CRC), regional relative bias, and relative variance.

Results

The bias and variance of the phantom brain image reconstructed using the patch-based method were 0.421 and 5.035, respectively, and this process took 83.637 seconds. The bias and variance of the image reconstructed by the proposed improved method, however, were 0.396 and 4.568, respectively, and this process took 41.851 seconds. This demonstrates that the proposed algorithm accelerated the reconstruction convergence. The CRC of the phantom brain image reconstructed using the patch-based method was iterated 20 times and reached 0.284, compared with the proposed method, which reached 0.446. When using a count of 5,000 K data obtained from the mouse study, both the patch-based method and the proposed method reconstructed images similar to the ground truth image. The intensity of the ground truth image was 88.3, and it was located in the 102^nd row and the 116^th column. However, when the count was reduced to below 40 K, and the patch-based method was used, image quality was significantly reduced. This effect was not observed when the proposed method was used. When a count of 40 K was used, the image intensity was 58.79 when iterated 100 times by the patch-based method, and it was located in the 102^nd row and the 116^th column, while the intensity when iterated 50 times by the proposed method was 63.83. This suggests that the proposed method improves image reconstruction from low-count data.

Conclusions

This improved method of PET image reconstruction could potentially improve the quality of PET images faster than other methods and also produce better reconstructions from low-count data.

Neural MLAA for PET-enabled Dual-Energy CT Imaging

The PET-enabled dual-energy CT method allows dual-energy CT imaging on PET/CT scanners without the need for a second x-ray CT scan. A 511 keV γ-ray attenuation image can be reconstructed from time-of-flight PET emission data using the maximum-likelihood attenuation and activity (MLAA) algorithm. However, the attenuation image reconstructed by standard MLAA is commonly noisy. To suppress noise, we propose a neural-network approach for MLAA reconstruction. The PET attenuation image is described as a function of kernelized neural networks with the flexibility of incorporating the available x-ray CT image as anatomical prior. By applying optimization transfer, the complex optimization problem of the proposed neural MLAA reconstruction is solved by a modularized iterative algorithm with each iteration decomposed into three steps: PET activity image update, attenuation image update, and neural-network learning using a weighed mean squared-error loss. The optimization algorithm is guaranteed to monotonically increase the data likelihood. The results from computer simulations demonstrated the neural MLAA algorithm can achieve a significant improvement on the γ-ray CT image quality as compared to other algorithms.

4D deep image prior: dynamic PET image denoising using an unsupervised four-dimensional branch convolutional neural network

Although convolutional neural networks (CNNs) demonstrate the superior performance in denoising positron emission tomography (PET) images, a supervised training of the CNN requires a pair of large, high-quality PET image datasets. As an unsupervised learning method, a deep image prior (DIP) has recently been proposed; it can perform denoising with only the target image. In this study, we propose an innovative procedure for the DIP approach with a four-dimensional (4D) branch CNN architecture in end-to-end training to denoise dynamic PET images. Our proposed 4D CNN architecture can be applied to end-to-end dynamic PET image denoising by introducing a feature extractor and a reconstruction branch for each time frame of the dynamic PET image. In the proposed DIP method, it is not necessary to prepare high-quality and large patient-related PET images. Instead, a subject's own static PET image is used as additional information, dynamic PET images are treated as training labels, and denoised dynamic PET images are obtained from the CNN outputs. Both simulation with [¹⁸F]fluoro-2-deoxy-D-glucose (FDG) and preclinical data with [¹⁸F]FDG and [¹¹C]raclopride were used to evaluate the proposed framework. The results showed that our 4D DIP framework quantitatively and qualitatively outperformed 3D DIP and other unsupervised denoising methods. The proposed 4D DIP framework thus provides a promising procedure for dynamic PET image denoising.

3D/4D Reconstruction and Quantitative Total Body Imaging

Total-body PET image reconstruction follows a similar procedure to the image reconstruction process for standard whole-body PET scanners. One unique aspect of total-body imaging is simultaneous coverage of the entire human body, which makes it convenient to perform total-body dynamic PET scans. Therefore, four-dimensional dynamic PET reconstruction and parametric imaging are of great interest in total-body imaging. This article covers some basics of PET image reconstruction and then focuses on three- and four-dimensional PET reconstruction for total-body imaging. Methods for image formation from raw measurements in total-body PET are described. Challenges and opportunities in total-body PET image reconstruction are discussed.

An iterative reconstruction method for sparse-projection data for low-dose CT

Reducing X-ray radiation is beneficial for reducing the risk of cancer in patients. There are two main approaches for achieving this goal namely, one is to reduce the X-ray current, and another is to apply sparse-view protocols to do image scanning and projections. However, these techniques usually lead to degradation of the reconstructed image quality, resulting in excessive noise and severe edge artifacts, which seriously affect the diagnosis result. In order to overcome such limitation, this study proposes and tests an algorithm based on guided kernel filtering. The algorithm combines the characteristics of anisotropic edges between adjacent image voxels, expresses the relevant weights with an exponential function, and adjusts the weights adaptively through local gray gradients to better preserve the image structure while suppressing noise information. Experiments show that the proposed method can effectively suppress noise and preserve the image structure. Comparing with similar algorithms, the proposed algorithm greatly improves the peak signal-to-noise ratio (PSNR), structural similarity (SSIM), and root mean square error (RMSE) of the reconstructed image. The proposed algorithm has the best effect in quantitative analysis, which verifies the effectiveness of the proposed method and good image reconstruction performance. Overall, this study demonstrates that the proposed method can reduce the number of projections required for repeated CT scans and has potential for medical applications in reducing radiation doses.

RetroSPECT: Gallium-67 as a Long-Lived Imaging Agent for Theranostics

A limitation to the wider introduction of personalised dosimetry in theranostics is the relative paucity of imaging radionuclides with suitable physical and chemical properties to be paired with a long-lived therapeutic partner. As most of the beta-emitting therapeutic radionuclides emit gamma radiation as well they could potentially be used as the imaging radionuclide as well as the therapeutic radionuclide. However, the downsides are that the beta radiation will deliver a significant radiation dose as part of the treatment planning procedure, and the gamma radiation branching ratio is often quite low. Gallium-67 has been in use in nuclear medicine for over 50 years. However, the tremendous interest in gallium imaging in theranostics in recent times has focused on the PET radionuclide gallium-68. In this article it is suggested that the longer-lived gallium-67, which has desirable characteristics for imaging with the gamma camera and a suitably long half-life to match biological timescales for drug uptake and turnover, has been overlooked, in particular, for treatment planning with radionuclide therapy. Gallium-67 could also allow non-PET facilities to participate in theranostic imaging prior to treatment or for monitoring response after therapy. Gallium-67 could play a niche role in the future development of personalised medicine with theranostics.

PET-enabled dual-energy CT: image reconstruction and a proof-of-concept computer simulation study

Standard dual-energy computed tomography (CT) uses two different x-ray energies to obtain energy-dependent tissue attenuation information to allow quantitative material decomposition. The combined use of dual-energy CT and positron emission tomography (PET) may provide a more comprehensive characterization of disease states in cancer and other diseases. However, the integration of dual-energy CT with PET is not trivial, either requiring costly hardware upgrades or increasing radiation exposure. This paper proposes a different dual-energy CT imaging method that is enabled by PET. Instead of using a second x-ray CT scan with a different energy, this method exploits time-of-flight PET image reconstruction via the maximum likelihood attenuation and activity (MLAA) algorithm to obtain a 511 keV gamma-ray attenuation image from PET emission data. The high-energy gamma-ray attenuation image is then combined with the low-energy x-ray CT of PET/CT to provide a pair of dual-energy CT images. A major challenge with the standard MLAA reconstruction is the high noise present in the reconstructed 511 keV attenuation map, which would not compromise the PET activity reconstruction too much but may significantly affect the performance of the gamma-ray attenuation image for material decomposition. To overcome the problem, we further propose a kernel MLAA algorithm to exploit the prior information from the available x-ray CT image. We conducted a computer simulation to test the concept and algorithm for the task of material decomposition. The simulation results demonstrate that this PET-enabled dual-energy CT method is promising for quantitative material decomposition. The proposed method can be readily implemented on time-of-flight PET/CT scanners to enable simultaneous PET and dual-energy CT imaging.

Penalized-likelihood PET Image Reconstruction Using 3D Structural Convolutional Sparse Coding

Positron emission tomography (PET) is widely used for clinical diagnosis. As PET suffers from low resolution and high noise, numerous efforts try to incorporate anatomical priors into PET image reconstruction, especially with the development of hybrid PET/CT and PET/MRI systems. In this work, we proposed a cube-based 3D structural convolutional sparse coding (CSC) concept for penalized-likelihood PET image reconstruction, named 3D PET-CSC. The proposed 3D PET-CSC takes advantage of the convolutional operation and manages to incorporate anatomical priors without the need of registration or supervised training. As 3D PET-CSC codes the whole 3D PET image, instead of patches, it alleviates the staircase artifacts commonly presented in traditional patch-based sparse coding methods. Compared with traditional coding methods in Fourier domain, the proposed method extends the 3D CSC to a straightforward approach based on the pursuit of localized cubes. Moreover, we developed the residual-image and order-subset mechanisms to further reduce the computational cost and accelerate the convergence for the proposed 3D PET-CSC method. Experiments based on computer simulations and clinical datasets demonstrate the superiority of 3D PET-CSC compared with other reference methods.

Kernel-based curve-fitting method with spatial regularization for generation of parametric images in dynamic PET

Due to high levels of noise in pixel-wise time-activity curves, the indirect method that generates kinetic parametric images from dynamic positron emission tomography (PET) images often results in poor parametric image quality. We have demonstrated that the quality of parametric images can be improved by denoising dynamic PET images, using gradient-free curve-fitting and applying a kernel-based post-filtering to parametric images. However, many gradient-free curve-fitting methods are time-consuming. Moreover, some parameter estimates (e.g. k₂ and k₃) have large variability. To provide high-quality PET parametric images with low computational cost, we propose a curve-fitting method that incorporates the kernel-based denoising method and the highly constrained backprojection technique into the Levenberg-Marquardt (LM) algorithm. We conducted a simulation study to evaluate the performance of the proposed curve-fitting method. Dynamic PET images were reconstructed using the expectation-maximization (EM) algorithm and were denoised before parameter estimation. Compared to the LM algorithm with and without the kernel-based post-filtering, the proposed method achieved superior performance, offering a decrease in both bias and coefficient of variation (CV) on all parametric images. Overall, the proposed method exhibited lower bias and slightly higher CV than the gradient-free pattern search method with the kernel-based post-filtering (PatS-K). Moreover, the computation time of the proposed method was about 18 times lower than that of the PatS-K method. Finally, we show that the proposed method can further improve the quality of parametric images when dynamic PET images are reconstructed using the kernel-based EM algorithm.

PET Parametric Imaging: Past, Present, and Future

Positron emission tomography (PET) is actively used in a diverse range of applications in oncology, cardiology, and neurology. The use of PET in the clinical setting focuses on static (single time frame) imaging at a specific time-point post radiotracer injection and is typically considered as semi-quantitative; e.g. standardized uptake value (SUV) measures. In contrast, dynamic PET imaging requires increased acquisition times but has the advantage that it measures the full spatiotemporal distribution of a radiotracer and, in combination with tracer kinetic modeling, enables the generation of multiparametric images that more directly quantify underlying biological parameters of interest, such as blood flow, glucose metabolism, and receptor binding. Parametric images have the potential for improved detection and for more accurate and earlier therapeutic response assessment. Parametric imaging with dynamic PET has witnessed extensive research in the past four decades. In this paper, we provide an overview of past and present activities and discuss emerging opportunities in the field of parametric imaging for the future.

Multiparametric Cardiac 18F-FDG PET in Humans: Kinetic Model Selection and Identifiability Analysis

Cardiac ¹⁸F-FDG PET has been used in clinics to assess myocardial glucose metabolism. Its ability for imaging myocardial glucose transport, however, has rarely been exploited in clinics. Using the dynamic FDG-PET scans of ten patients with coronary artery disease, we investigate in this paper appropriate dynamic scan and kinetic modeling protocols for efficient quantification of myocardial glucose transport. Three kinetic models and the effect of scan duration were evaluated by using statistical fit quality, assessing the impact on kinetic quantification, and analyzing the practical identifiability. The results show that the kinetic model selection depends on the scan duration. The reversible two-tissue model was needed for a one-hour dynamic scan. The irreversible two-tissue model was optimal for a scan duration of around 10-15 minutes. If the scan duration was shortened to 2-3 minutes, a one-tissue model was the most appropriate. For global quantification of myocardial glucose transport, we demonstrated that an early dynamic scan with a duration of 10-15 minutes and irreversible kinetic modeling was comparable to the full one-hour scan with reversible kinetic modeling. Myocardial glucose transport quantification provides an additional physiological parameter on top of the existing assessment of glucose metabolism and has the potential to enable single tracer multiparametric imaging in the myocardium.

Applications of artificial intelligence and deep learning in molecular imaging and radiotherapy

This brief review summarizes the major applications of artificial intelligence (AI), in particular deep learning approaches, in molecular imaging and radiation therapy research. To this end, the applications of artificial intelligence in five generic fields of molecular imaging and radiation therapy, including PET instrumentation design, PET image reconstruction quantification and segmentation, image denoising (low-dose imaging), radiation dosimetry and computer-aided diagnosis, and outcome prediction are discussed. This review sets out to cover briefly the fundamental concepts of AI and deep learning followed by a presentation of seminal achievements and the challenges facing their adoption in clinical setting.

Generative adversarial network based regularized image reconstruction for PET

Positron emission tomography (PET) is an ill-posed inverse problem and suffers high noise due to limited number of detected events. Prior information can be used to improve the quality of reconstructed PET images. Deep neural networks have also been applied to regularized image reconstruction. One method is to use a pretrained denoising neural network to represent the PET image and to perform a constrained maximum likelihood estimation. In this work, we propose to use a generative adversarial network (GAN) to further improve the network performance. We also modify the objective function to include a data-matching term on the network input. Experimental studies using computer-based Monte Carlo simulations and real patient datasets demonstrate that the proposed method leads to noticeable improvements over the kernel-based and U-net-based regularization methods in terms of lesion contrast recovery versus background noise trade-offs.

Bootstrap-Optimised Regularised Image Reconstruction for Emission Tomography

In emission tomography, iterative image reconstruction from noisy measured data usually results in noisy images, and so regularisation is often used to compensate for noise. However, in practice, an appropriate, automatic and precise specification of the strength of regularisation for image reconstruction from a given noisy measured dataset remains unresolved. Existing approaches are either empirical approximations with no guarantee of generalisation, or else are computationally intensive cross-validation methods requiring full reconstructions for a limited set of preselected regularisation strengths. In contrast, we propose a novel methodology embedded within iterative image reconstruction, using one or more bootstrapped replicates of the measured data for precise optimisation of the regularisation. The approach uses a conventional unregularised iterative update of a current image estimate from the noisy measured data, and then also uses the bootstrap replicate to obtain a bootstrap update of the current image estimate. The method then seeks the regularisation hyperparameters which, when applied to the bootstrap update of the image, lead to a best fit of the regularised bootstrap update to the conventional measured data update. This corresponds to estimating the degree of regularisation needed in order to map the noisy update to a model of the mean of an ensemble of noisy updates. For a given regularised objective function (e.g. penalised likelihood), no hyperparameter selection or tuning is required. The method is demonstrated for positron emission tomography (PET) data at different noise levels, and delivers near-optimal reconstructions (in terms of reconstruction error) without any knowledge of the ground truth, nor any form of training data.

Subsecond total-body imaging using ultrasensitive positron emission tomography

A 194-cm-long total-body positron emission tomography/computed tomography (PET/CT) scanner (uEXPLORER), has been constructed to offer a transformative platform for human radiotracer imaging in clinical research and healthcare. Its total-body coverage and exceptional sensitivity provide opportunities for innovative studies of physiology, biochemistry, and pharmacology. The objective of this study is to develop a method to perform ultrahigh (100 ms) temporal resolution dynamic PET imaging by combining advanced dynamic image reconstruction paradigms with the uEXPLORER scanner. We aim to capture the fast dynamics of initial radiotracer distribution, as well as cardiac motion, in the human body. The results show that we can visualize radiotracer transport in the body on timescales of 100 ms and obtain motion-frozen images with superior image quality compared to conventional methods. The proposed method has applications in studying fast tracer dynamics, such as blood flow and the dynamic response to neural modulation, as well as performing real-time motion tracking (e.g., cardiac and respiratory motion, and gross body motion) without any external monitoring device (e.g., electrocardiogram, breathing belt, or optical trackers).

PET image reconstruction using physical and mathematical modelling for time of flight PET-MR scanners in the STIR library

This work demonstrates how computational and physical modelling of the positron emission tomography (PET) image acquisition process for a state-of-the-art integrated PET and magnetic resonance imaging (PET-MR) system can produce images comparable to the manufacturer. The GE SIGNA PET/MR scanner is manufactured by General Electric and has time-of-flight (TOF) capabilities of about 390 ps. All software development took place in the Software for Tomographic Image Reconstruction (STIR: http://stir.sf.net) library, which is a widely used open source software to reconstruct data as exported from emission tomography scanners. The new software developments will be integrated into STIR, providing the opportunity for researchers worldwide to establish and expand their image reconstruction methods. Furthermore, this work is of particular significance as it provides the first validation of TOF PET image reconstruction for real scanner datasets using the STIR library. This paper presents the methodology, analysis, and critical issues encountered in implementing an independent reconstruction software package. Acquired PET data were processed via several appropriate algorithms which are necessary to produce an accurate and precise quantitative image. This included mathematical, physical and anatomical modelling of the patient and simulation of various aspects of the acquisition. These included modelling of random coincidences using 'singles' rates per crystals, detector efficiencies and geometric effects. Attenuation effects were calculated by using the STIR's attenuation correction model. Modelling all these effects within the system matrix allowed the reconstruction of PET images which demonstrates the metabolic uptake of the administered radiopharmaceutical. These implementations were validated using measured phantom and clinical datasets. The developments are tested using the ordered subset expectation maximisation (OSEM) and the more recently proposed kernelised expectation maximisation (KEM) algorithm which incorporates anatomical information from MR images into PET reconstruction.

Kernel-Based Reconstruction of C-11-Hydroxyephedrine Cardiac PET Images of the Sympathetic Nervous System

Image reconstruction for positron emission tomography (PET) can be challenging and the resulting image typically has high noise. The kernel-based reconstruction method [1], incorporates prior anatomic information in the reconstruction algorithm to reduce noise while preserving resolution. Prior information is incorporated in the reconstruction algorithm by means of spatial kernels originally used in machine learning. In this paper, the kernel-based method is used to reconstruct PET images of sympathetic innervation in the heart. The resulting images are compared with standard Ordered Subset Expectation Maximization (OSEM) reconstructed images qualitatively and quantitatively using data from 6 human subjects. The kernel-based method demonstrated superior SNR with preserved contrast and accuracy compared to OSEM.

Machine Learning in PET: from Photon Detection to Quantitative Image Reconstruction

Machine learning has found unique applications in nuclear medicine from photon detection to quantitative image reconstruction. While there have been impressive strides in detector development for time-of-flight positron emission tomography, most detectors still make use of simple signal processing methods to extract the time and position information from the detector signals. Now with the availability of fast waveform digitizers, machine learning techniques have been applied to estimate the position and arrival time of high-energy photons. In quantitative image reconstruction, machine learning has been used to estimate various corrections factors, including scattered events and attenuation images, as well as to reduce statistical noise in reconstructed images. Here machine learning either provides a faster alternative to an existing time-consuming computation, such as in the case of scatter estimation, or creates a data-driven approach to map an implicitly defined function, such as in the case of estimating the attenuation map for PET/MR scans. In this article, we will review the abovementioned applications of machine learning in nuclear medicine.

Low-dose CT reconstruction method based on prior information of normal-dose image

BACKGROUND

Radiation risk from computed tomography (CT) is always an issue for patients, especially those in clinical conditions in which repeated CT scanning is required. For patients undergoing repeated CT scanning, a low-dose protocol, such as sparse scanning, is often used, and consequently, an advanced reconstruction algorithm is also needed.

OBJECTIVE

To develop a novel algorithm used for sparse-view CT reconstruction associated with the prior image.

METHODS

A low-dose CT reconstruction method based on prior information of normal-dose image (PI-NDI) involving a transformed model for attenuation coefficients of the object to be reconstructed and prior information application in the forward-projection process was used to reconstruct CT images from sparse-view projection data. A digital extended cardiac-torso (XCAT) ventral phantom and a diagnostic head phantom were employed to evaluate the performance of the proposed PI-NDI method. The root-mean-square error (RMSE), peak signal-to-noise ratio (PSNR) and mean percent absolute error (MPAE) of the reconstructed images were measured for quantitative evaluation of the proposed PI-NDI method.

RESULTS

The reconstructed images with sparse-view projection data via the proposed PI-NDI method have higher quality by visual inspection than that via the compared methods. In terms of quantitative evaluations, the RMSE measured on the images reconstructed by the PI-NDI method with sparse projection data is comparable to that by MLEM-TV, PWLS-TV and PWLS-PICCS with fully sampled projection data. When the projection data are very sparse, images reconstructed by the PI-NDI method have higher PSNR values and lower MPAE values than those from the compared algorithms.

CONCLUSIONS

This study presents a new low-dose CT reconstruction method based on prior information of normal-dose image (PI-NDI) for sparse-view CT image reconstruction. The experimental results validate that the new method has superior performance over other state-of-art methods.

3D Tensor Based Nonlocal Low Rank Approximation in Dynamic PET Reconstruction

Reconstructing images from multi-view projections is a crucial task both in the computer vision community and in the medical imaging community, and dynamic positron emission tomography (PET) is no exception. Unfortunately, image quality is inevitably degraded by the limitations of photon emissions and the trade-off between temporal and spatial resolution. In this paper, we develop a novel tensor based nonlocal low-rank framework for dynamic PET reconstruction. Spatial structures are effectively enhanced not only by nonlocal and sparse features, but momentarily by tensor-formed low-rank approximations in the temporal realm. Moreover, the total variation is well regularized as a complementation for denoising. These regularizations are efficiently combined into a Poisson PET model and jointly solved by distributed optimization. The experiments demonstrated in this paper validate the excellent performance of the proposed method in dynamic PET.

Intercomparison of MR-informed PET image reconstruction methods

PURPOSE

Numerous image reconstruction methodologies for positron emission tomography (PET) have been developed that incorporate magnetic resonance (MR) imaging structural information, producing reconstructed images with improved suppression of noise and reduced partial volume effects. However, the influence of MR structural information also increases the possibility of suppression or bias of structures present only in the PET data (PET-unique regions). To address this, further developments for MR-informed methods have been proposed, for example, through inclusion of the current reconstructed PET image, alongside the MR image, in the iterative reconstruction process. In this present work, a number of kernel and maximum a posteriori (MAP) methodologies are compared, with the aim of identifying methods that enable a favorable trade-off between the suppression of noise and the retention of unique features present in the PET data.

METHODS

The reconstruction methods investigated were: the MR-informed conventional and spatially compact kernel methods, referred to as KEM and KEM largest value sparsification (LVS) respectively; the MR-informed Bowsher and Gaussian MR-guided MAP methods; and the PET-MR-informed hybrid kernel and anato-functional MAP methods. The trade-off between improving the reconstruction of the whole brain region and the PET-unique regions was investigated for all methods in comparison with postsmoothed maximum likelihood expectation maximization (MLEM), evaluated in terms of structural similarity index (SSIM), normalized root mean square error (NRMSE), bias, and standard deviation. Both simulated BrainWeb (10 noise realizations) and real [18 F] fluorodeoxyglucose (FDG) three-dimensional datasets were used. The real [18 F]FDG dataset was augmented with simulated tumors to allow comparison of the reconstruction methodologies for the case of known regions of PET-MR discrepancy and evaluated at full counts (100%) and at a reduced (10%) count level.

RESULTS

For the high-count simulated and real data studies, the anato-functional MAP method performed better than the other methods under investigation (MR-informed, PET-MR-informed and postsmoothed MLEM), in terms of achieving the best trade-off for the reconstruction of the whole brain and PET-unique regions, assessed in terms of the SSIM, NRMSE, and bias vs standard deviation. The inclusion of PET information in the anato-functional MAP method enables the reconstruction of PET-unique regions to attain similarly low levels of bias as unsmoothed MLEM, while moderately improving the whole brain image quality for low levels of regularization. However, for low count simulated datasets the anato-functional MAP method performs poorly, due to the inclusion of noisy PET information in the regularization term. For the low counts simulated dataset, KEM LVS and to a lesser extent, HKEM performed better than the other methods under investigation in terms of achieving the best trade-off for the reconstruction of the whole brain and PET-unique regions, assessed in terms of the SSIM, NRMSE, and bias vs standard deviation.

CONCLUSION

For the reconstruction of noisy data, multiple MR-informed methods produce favorable whole brain vs PET-unique region trade-off in terms of the image quality metrics of SSIM and NRMSE, comfortably outperforming the whole image denoising of postsmoothed MLEM.

MRI-Driven PET Image Optimization for Neurological Applications

Positron emission tomography (PET) and magnetic resonance imaging (MRI) are established imaging modalities for the study of neurological disorders, such as epilepsy, dementia, psychiatric disorders and so on. Since these two available modalities vary in imaging principle and physical performance, each technique has its own advantages and disadvantages over the other. To acquire the mutual complementary information and reinforce each other, there is a need for the fusion of PET and MRI. This combined dual-modality (either sequential or simultaneous) could generate preferable soft tissue contrast of brain tissue, flexible acquisition parameters, and minimized exposure to radiation. The most unique superiority of PET/MRI is mainly manifested in MRI-based improvement for the inherent limitations of PET, such as motion artifacts, partial volume effect (PVE) and invasive procedure in quantitative analysis. Head motion during scanning significantly deteriorates the effective resolution of PET image, especially for the dynamic scan with lengthy time. Hybrid PET/MRI device can offer motion correction (MC) for PET data through MRI information acquired simultaneously. Regarding the PVE associated with limited spatial resolution, the process and reconstruction of PET data can be further optimized by using acquired MRI either sequentially or simultaneously. The quantitative analysis of dynamic PET data mainly relies upon an invasive arterial blood sampling procedure to acquire arterial input function (AIF). An image-derived input function (IDIF) method without the need of arterial cannulization, can serve as a potential alternative estimation of AIF. Compared with using PET data only, combining anatomical or functional information from MRI for improving the accuracy in IDIF approach has been demonstrated. Yet, due to the interference and inherent disparity between the two modalities, these methods for optimizing PET image based on MRI still have many technical challenges. This review discussed upon the most recent progress, current challenges and future directions of MRI-driven PET data optimization for neurological applications, with either sequential or simultaneous acquisition approach.

PET Image Reconstruction Using Deep Image Prior

Recently, deep neural networks have been widely and successfully applied in computer vision tasks and have attracted growing interest in medical imaging. One barrier for the application of deep neural networks to medical imaging is the need for large amounts of prior training pairs, which is not always feasible in clinical practice. This is especially true for medical image reconstruction problems, where raw data are needed. Inspired by the deep image prior framework, in this paper, we proposed a personalized network training method where no prior training pairs are needed, but only the patient's own prior information. The network is updated during the iterative reconstruction process using the patient-specific prior information and measured data. We formulated the maximum-likelihood estimation as a constrained optimization problem and solved it using the alternating direction method of multipliers algorithm. Magnetic resonance imaging guided positron emission tomography reconstruction was employed as an example to demonstrate the effectiveness of the proposed framework. Quantification results based on simulation and real data show that the proposed reconstruction framework can outperform Gaussian post-smoothing and anatomically guided reconstructions using the kernel method or the neural-network penalty.

Personalized Models for Injected Activity Levels in SPECT Myocardial Perfusion Imaging

We propose a patient-specific ("personalized") approach for tailoring the injected activities to individual patients in order to achieve dose reduction in SPECT-myocardial perfusion imaging (MPI). First, we develop a strategy to determine the minimum dose levels required for each patient in a large set of clinical acquisitions (857 subjects) such that the reconstructed images are sufficiently similar to that obtained at conventional clinical dose. We then apply machine learning models to predict the required dose levels on an individual basis based on a set of patient attributes which include body measurements and various clinical variables. We demonstrate the personalized dose models for two commonly used reconstruction methods in clinical SPECT-MPI: 1) conventional filtered backprojection (FBP) with post-filtering and 2) ordered-subsets expectation-maximization (OS-EM) with corrections for attenuation, scatter and resolution, and evaluate their performance in perfusion-defect detection by using the clinical Quantitative Perfusion SPECT software package. The results indicate that the achieved dose reduction can vary greatly among individuals from their conventional clinical dose and that the personalized dose models can achieve further reduction on average compared with a global (non-patient specific) dose reduction approach. In particular, the average personalized dose level can be reduced to 58% and 54% of the full clinical dose, respectively, for FBP and OS-EM reconstruction, while without deteriorating the accuracy in perfusion-defect detection. Furthermore, with the average personalized dose further reduced to only 16% of full dose, OS-EM can still achieve a detection accuracy level comparable to that of FBP with full dose.

Higher SNR PET image prediction using a deep learning model and MRI image

PET images often suffer poor signal-to-noise ratio (SNR). Our objective is to improve the SNR of PET images using a deep neural network (DNN) model and MRI images without requiring any higher SNR PET images in training. Our proposed DNN model consists of three modified U-Nets (3U-net). The PET training input data and targets were reconstructed using filtered-backprojection (FBP) and maximum likelihood expectation maximization (MLEM), respectively. FBP reconstruction was used because of its computational efficiency so that the trained network not only removes noise, but also accelerates image reconstruction. Digital brain phantoms downloaded from BrainWeb were used to evaluate the proposed method. Poisson noise was added into sinogram data to simulate a 6 min brain PET scan. Attenuation effect was included and corrected before the image reconstruction. Extra Poisson noise was introduced to the training inputs to improve the network denoising capability. Three independent experiments were conducted to examine the reproducibility. A lesion was inserted into testing data to evaluate the impact of mismatched MRI information using the contrast-to-noise ratio (CNR). The negative impact on noise reduction was also studied when miscoregistration between PET and MRI images occurs. Compared with 1U-net trained with only PET images, training with PET/MRI decreased the mean squared error (MSE) by 31.3% and 34.0% for 1U-net and 3U-net, respectively. The MSE reduction is equivalent to an increase in the count level by 2.5 folds and 2.9 folds for 1U-net and 3U-net, respectively. Compared with the MLEM images, the lesion CNR was improved 2.7 folds and 1.4 folds for 1U-net and 3U-net, respectively. The results show that the proposed method could improve the PET SNR without having higher SNR PET images.