Super-Resolution PET Imaging Using Convolutional Neural Networks

Anatomically Guided PET Image Reconstruction Using Conditional Weakly-Supervised Multi-Task Learning Integrating Self-Attention

To address the lack of high-quality training labels in positron emission tomography (PET) imaging, weakly-supervised reconstruction methods that generate network-based mappings between prior images and noisy targets have been developed. However, the learned model has an intrinsic variance proportional to the average variance of the target image. To suppress noise and improve the accuracy and generalizability of the learned model, we propose a conditional weakly-supervised multi-task learning (MTL) strategy, in which an auxiliary task is introduced serving as an anatomical regularizer for the PET reconstruction main task. In the proposed MTL approach, we devise a novel multi-channel self-attention (MCSA) module that helps learn an optimal combination of shared and task-specific features by capturing both local and global channel-spatial dependencies. The proposed reconstruction method was evaluated on NEMA phantom PET datasets acquired at different positions in a PET/CT scanner and 26 clinical whole-body PET datasets. The phantom results demonstrate that our method outperforms state-of-the-art learning-free and weakly-supervised approaches obtaining the best noise/contrast tradeoff with a significant noise reduction of approximately 50.0% relative to the maximum likelihood (ML) reconstruction. The patient study results demonstrate that our method achieves the largest noise reductions of 67.3% and 35.5% in the liver and lung, respectively, as well as consistently small biases in 8 tumors with various volumes and intensities. In addition, network visualization reveals that adding the auxiliary task introduces more anatomical information into PET reconstruction than adding only the anatomical loss, and the developed MCSA can abstract features and retain PET image details.

Virtual cylindrical PET for efficient DOI image reconstruction with sub-millimetre resolution

Objective.Image reconstruction in high resolution, narrow bore PET scanners with depth of interaction (DOI) capability presents a substantial computational challenge due to the very high sampling in detector and image space. The aim of this study is to evaluate the use of a virtual cylinder in reducing the number of lines of response (LOR) for DOI-based reconstruction in high resolution PET systems while maintaining uniform sub-millimetre spatial resolution.Approach.Virtual geometry was investigated using the awake animal mousePET as a high resolution test case. Using GEANT4 Application for Tomographic Emission (GATE), we simulated the physical scanner and three virtual cylinder implementations with detector size 0.74 mm, 0.47 mm and 0.36 mm (vPET1, vPET2 and vPET3, respectively). The virtual cylinder condenses physical LORs stemming from various crystal pairs and DOI combinations, and which intersect a single virtual detector pair, into a single virtual LOR. Quantitative comparisons of the point spread function (PSF) at various positions within the field of view (FOV) were compared for reconstructions based on the vPET implementations and the physical scanner. We also assessed the impact of the anisotropic PSFs by reconstructing images of a micro Derenzo phantom.Main results.All virtual cylinder implementations achieved LOR data compression of at least 50% for DOI PET reconstruction. PSF anisotropy in radial and tangential profiles was chiefly influenced by DOI resolution and only marginally by virtual detector size. Spatial degradation introduced by virtual cylinders was most prominent in the axial profile. All virtual cylinders achieved sub-millimetre volumetric resolution across the FOV when 6-bin DOI reconstructions (3.3 mm DOI resolution) were performed. Using vPET2 with 6 DOI bins yielded nearly identical reconstructions to the non-virtual case in the transaxial plane, with an LOR compression factor of 86%. Resolution modelling significantly reduced the effects of the asymmetric PSF arising from the non-cylindrical geometry of mousePET.Significance.Narrow bore and high resolution PET scanners require detectors with DOI capability, leading to computationally demanding reconstructions due to the large number of LORs. In this study, we show that DOI PET reconstruction with 50%-86% LOR compression is possible using virtual cylinders while maintaining sub-millimetre spatial resolution throughout the FOV. The methodology and analysis can be extended to other scanners with DOI capability intended for high resolution PET imaging.

Super-resolution techniques for biomedical applications and challenges

Super-resolution (SR) techniques have revolutionized the field of biomedical applications by detailing the structures at resolutions beyond the limits of imaging or measuring tools. These techniques have been applied in various biomedical applications, including microscopy, magnetic resonance imaging (MRI), computed tomography (CT), X-ray, electroencephalogram (EEG), ultrasound, etc. SR methods are categorized into two main types: traditional non-learning-based methods and modern learning-based approaches. In both applications, SR methodologies have been effectively utilized on biomedical images, enhancing the visualization of complex biological structures. Additionally, these methods have been employed on biomedical data, leading to improvements in computational precision and efficiency for biomedical simulations. The use of SR techniques has resulted in more detailed and accurate analyses in diagnostics and research, essential for early disease detection and treatment planning. However, challenges such as computational demands, data interpretation complexities, and the lack of unified high-quality data persist. The article emphasizes these issues, underscoring the need for ongoing development in SR technologies to further improve biomedical research and patient care outcomes.

3D-MRI super-resolution reconstruction using multi-modality based on multi-resolution CNN

BACKGROUND AND OBJECTIVE

High-resolution (HR) MR images provide rich structural detail to assist physicians in clinical diagnosis and treatment plan. However, it is arduous to acquire HR MRI due to equipment limitations, scanning time or patient comfort. Instead, HR MRI could be obtained through a number of computer assisted post-processing methods that have proven to be effective and reliable. This paper aims to develop a convolutional neural network (CNN) based super-resolution reconstruction framework for low-resolution (LR) T2w images.

METHOD

In this paper, we propose a novel multi-modal HR MRI generation framework based on deep learning techniques. Specifically, we construct a CNN based on multi-resolution analysis to learn an end-to-end mapping between LR T2w and HR T2w, where HR T1w is fed into the network to offer detailed a priori information to help generate HR T2w. Furthermore, a low-frequency filtering module is introduced to filter out the interference from HR-T1w during high-frequency information extraction. Based on the idea of multi-resolution analysis, detailed features extracted from HR T1w and LR T2w are fused at two scales in the network and then HR T2w is reconstructed by upsampling and dense connectivity module.

RESULTS

Extensive quantitative and qualitative evaluations demonstrate that the proposed method enhances the recovered HR T2w details and outperforms other state-of-the-art methods. In addition, the experimental results also suggest that our network has a lightweight structure and favorable generalization performance.

CONCLUSION

The results show that the proposed method is capable of reconstructing HR T2w with higher accuracy. Meanwhile, the super-resolution reconstruction results on other dataset illustrate the excellent generalization ability of the method.

Impact of using sinogram domain data in the super-resolution of CT images on diagnostic information

BACKGROUND

In recent times, deep-learning-based super-resolution (DL-SR) techniques for computed tomography (CT) images have shown outstanding results in terms of full-reference image quality (FR-IQ) metrics (e.g., root mean square error and structural similarity index metric), which assesses IQ by measuring its similarity to the high-resolution (HR) image. In addition, IQ can be evaluated via task-based IQ (Task-IQ) metrics that evaluate the ability to perform specific tasks. Ironically, most proposed image domain-based SR techniques are not possible to improve a Task-IQ metric, which assesses the amount of information related to diagnosis.

PURPOSE

In the case of CT imaging systems, sinogram domain data can be utilized for SR techniques. Therefore, this study aims to investigate the impact of utilizing sinogram domain data on diagnostic information restoration ability.

METHODS

We evaluated three DL-SR techniques: using image domain data (Image-SR), using sinogram domain data (Sinogram-SR), and using sinogram as well as image domain data (Dual-SR). For Task-IQ evaluation, the Rayleigh discrimination task was used to evaluate diagnostic ability by focusing on the resolving power aspect, and an ideal observer (IO) can be used to perform the task. In this study, we used a convolutional neural network (CNN)-based IO that approximates the IO performance. We compared the IO performances of the SR techniques according to the data domain to evaluate the discriminative information restoration ability.

RESULTS

Overall, the low-resolution (LR) and SR exhibit lower IO performances compared with that of HR owing to their degraded discriminative information when detector binning is used. Next, between the SR techniques, Image-SR does not show superior IO performances compared to the LR image, but Sinogram-SR and Dual-SR show superior IO performances than the LR image. Furthermore, in Sinogram-SR, we confirm that FR-IQ and IO performance are positively correlated. These observations demonstrate that sinogram domain upsampling improves the representation ability for discriminative information in the image domain compared to the LR and Image-SR.

CONCLUSIONS

Unlike Image-SR, Sinogram-SR can improve the amount of discriminative information present in the image domain. This demonstrates that to improve the amount of discriminative information on the resolving power aspect, it is necessary to employ sinogram domain processing.

Quasi-supervised learning for super-resolution PET

Low resolution of positron emission tomography (PET) limits its diagnostic performance. Deep learning has been successfully applied to achieve super-resolution PET. However, commonly used supervised learning methods in this context require many pairs of low- and high-resolution (LR and HR) PET images. Although unsupervised learning utilizes unpaired images, the results are not as good as that obtained with supervised deep learning. In this paper, we propose a quasi-supervised learning method, which is a new type of weakly-supervised learning methods, to recover HR PET images from LR counterparts by leveraging similarity between unpaired LR and HR image patches. Specifically, LR image patches are taken from a patient as inputs, while the most similar HR patches from other patients are found as labels. The similarity between the matched HR and LR patches serves as a prior for network construction. Our proposed method can be implemented by designing a new network or modifying an existing network. As an example in this study, we have modified the cycle-consistent generative adversarial network (CycleGAN) for super-resolution PET. Our numerical and experimental results qualitatively and quantitatively show the merits of our method relative to the state-of-the-art methods. The code is publicly available at https://github.com/PigYang-ops/CycleGAN-QSDL.

Partial volume correction of PET image data using geometric transfer matrices based on uniform B-splines

Objective. Most methods for partial volume correction (PVC) of positron emission tomography (PET) data employ anatomical segmentation of images into regions of interest. This approach is not optimal for exploratory functional imaging beyond regional hypotheses. Here, we describe a novel method for unbiased voxel-wise PVC.Approach.B-spline basis functions were combined with geometric transfer matrices to enable a method (bsGTM) that provides PVC or alternatively provides smoothing with minimal regional crosstalk. The efficacy of the proposed method was evaluated using Monte Carlo simulations, human PET data, and murine functional PET data.Main results.In simulations, bsGTM provided recovery of partial volume signal loss comparable to iterative deconvolution, while demonstrating superior resilience to noise. In a real murine PET dataset, bsGTM yielded much higher sensitivity for detecting amphetamine-induced reduction of [11C]raclopride binding potential. In human PET data, bsGTM smoothing enabled increased signal-to-noise ratios with less degradation of binding potentials relative to Gaussian convolution or non-local means.Significance.bsGTM offers improved performance for PVC relative to iterative deconvolution, the current method of choice for voxel-wise PVC, especially in the common PET regime of low signal-to-noise ratio. The new method provides an anatomically unbiased way to compensate partial volume errors in cases where anatomical segmentation is unavailable or of questionable relevance or accuracy.

Artificial Intelligence for PET and SPECT Image Enhancement

Nuclear medicine imaging modalities such as PET and SPECT are confounded by high noise levels and low spatial resolution, necessitating postreconstruction image enhancement to improve their quality and quantitative accuracy. Artificial intelligence (AI) models such as convolutional neural networks, U-Nets, and generative adversarial networks have shown promising outcomes in enhancing PET and SPECT images. This review article presents a comprehensive survey of state-of-the-art AI methods for PET and SPECT image enhancement and seeks to identify emerging trends in this field. We focus on recent breakthroughs in AI-based PET and SPECT image denoising and deblurring. Supervised deep-learning models have shown great potential in reducing radiotracer dose and scan times without sacrificing image quality and diagnostic accuracy. However, the clinical utility of these methods is often limited by their need for paired clean and corrupt datasets for training. This has motivated research into unsupervised alternatives that can overcome this limitation by relying on only corrupt inputs or unpaired datasets to train models. This review highlights recently published supervised and unsupervised efforts toward AI-based PET and SPECT image enhancement. We discuss cross-scanner and cross-protocol training efforts, which can greatly enhance the clinical translatability of AI-based image enhancement tools. We also aim to address the looming question of whether the improvements in image quality generated by AI models lead to actual clinical benefit. To this end, we discuss works that have focused on task-specific objective clinical evaluation of AI models for image enhancement or incorporated clinical metrics into their loss functions to guide the image generation process. Finally, we discuss emerging research directions, which include the exploration of novel training paradigms, curation of larger task-specific datasets, and objective clinical evaluation that will enable the realization of the full translation potential of these models in the future.

SPECT reconstruction with a trained regularizer using CT-side information: Application to 177Lu SPECT imaging

Improving low-count SPECT can shorten scans and support pre-therapy theranostic imaging for dosimetry-based treatment planning, especially with radionuclides like 177Lu known for low photon yields. Conventional methods often underperform in low-count settings, highlighting the need for trained regularization in model-based image reconstruction. This paper introduces a trained regularizer for SPECT reconstruction that leverages segmentation based on CT imaging. The regularizer incorporates CT-side information via a segmentation mask from a pre-trained network (nnUNet). In this proof-of-concept study, we used patient studies with 177Lu DOTATATE to train and tested with phantom and patient datasets, simulating pre-therapy imaging conditions. Our results show that the proposed method outperforms both standard unregularized EM algorithms and conventional regularization with CT-side information. Specifically, our method achieved marked improvements in activity quantification, noise reduction, and root mean square error. The enhanced low-count SPECT approach has promising implications for theranostic imaging, post-therapy imaging, whole body SPECT, and reducing SPECT acquisition times.

Impact of brain segmentation methods on regional metabolism quantification in 18F-FDG PET/MR analysis

BACKGROUND

Accurate analysis of quantitative PET data plays a crucial role in studying small, specific brain structures. The integration of PET and MRI through an integrated PET/MR system presents an opportunity to leverage the benefits of precisely aligned structural MRI and molecular PET images in both spatial and temporal dimensions. However, in many clinical workflows, PET studies are often performed without the aid of individually matched structural MRI scans, primarily for the sake of convenience in the data collection and brain segmentation possesses. Currently, two commonly employed segmentation strategies for brain PET analysis are distinguished: methods with or without MRI registration and methods employing either atlas-based or individual-based algorithms. Moreover, the development of artificial intelligence (AI)-assisted methods for predicting brain segmentation holds promise but requires further validation of their efficiency and accuracy for clinical applications. This study aims to compare and evaluate the correlations, consistencies, and differences among the above-mentioned brain segmentation strategies in quantification of brain metabolism in 18F-FDG PET/MR analysis.

RESULTS

Strong correlations were observed among all methods (r = 0.932 to 0.999, P < 0.001). The variances attributable to subject and brain region were higher than those caused by segmentation methods (P < 0.001). However, intraclass correlation coefficient (ICC)s between methods with or without MRI registration ranged from 0.924 to 0.975, while ICCs between methods with atlas- or individual-based algorithms ranged from 0.741 to 0.879. Brain regions exhibiting significant standardized uptake values (SUV) differences due to segmentation methods were the basal ganglia nuclei (maximum to 11.50 ± 4.67%), and various cerebral cortexes in temporal and occipital regions (maximum to 18.03 ± 5.52%). The AI-based method demonstrated high correlation (r = 0.998 and 0.999, P < 0.001) and ICC (0.998 and 0.997) with FreeSurfer, substantially reducing the time from 8.13 h to 57 s on per subject.

CONCLUSIONS

Different segmentation methods may have impact on the calculation of brain metabolism in basal ganglia nuclei and specific cerebral cortexes. The AI-based approach offers improved efficiency and is recommended for its enhanced performance.

Medical image super-resolution reconstruction algorithms based on deep learning: A survey

BACKGROUND AND OBJECTIVE

With the high-resolution (HR) requirements of medical images in clinical practice, super-resolution (SR) reconstruction algorithms based on low-resolution (LR) medical images have become a research hotspot. This type of method can significantly improve image SR without improving hardware equipment, so it is of great significance to review it.

METHODS

Aiming at the unique SR reconstruction algorithms in the field of medical images, based on subdivided medical fields such as magnetic resonance (MR) images, computed tomography (CT) images, and ultrasound images. Firstly, we deeply analyzed the research progress of SR reconstruction algorithms, and summarized and compared the different types of algorithms. Secondly, we introduced the evaluation indicators corresponding to the SR reconstruction algorithms. Finally, we prospected the development trend of SR reconstruction technology in the medical field.

RESULTS

The medical image SR reconstruction technology based on deep learning can provide more abundant lesion information, relieve the expert's diagnosis pressure, and improve the diagnosis efficiency and accuracy.

CONCLUSION

The medical image SR reconstruction technology based on deep learning helps to improve the quality of medicine, provides help for the diagnosis of experts, and lays a solid foundation for the subsequent analysis and identification tasks of the computer, which is of great significance for improving the diagnosis efficiency of experts and realizing intelligent medical care.

Super-resolution in brain Positron Emission Tomography using a real-time motion capture system

Super-resolution (SR) is a methodology that seeks to improve image resolution by exploiting the increased spatial sampling information obtained from multiple acquisitions of the same target with accurately known sub-resolution shifts. This work aims to develop and evaluate an SR estimation framework for brain positron emission tomography (PET), taking advantage of a high-resolution infra-red tracking camera to measure shifts precisely and continuously. Moving phantoms and non-human primate (NHP) experiments were performed on a GE Discovery MI PET/CT scanner (GE Healthcare) using an NDI Polaris Vega (Northern Digital Inc), an external optical motion tracking device. To enable SR, a robust temporal and spatial calibration of the two devices was developed as well as a list-mode Ordered Subset Expectation Maximization (OSEM) PET reconstruction algorithm, incorporating the high-resolution tracking data from the Polaris Vega to correct motion for measured line of responses (LORs) on an event-by-event basis. For both phantoms and NHP studies, the SR reconstruction method yielded PET images with visibly increased spatial resolution compared to standard static acquisitions, allowing improved visualization of small structures. Quantitative analysis in terms of SSIM, CNR and line profiles were conducted and validated our observations. The results demonstrate that SR can be achieved in brain PET by measuring target motion in real-time using a high-resolution infrared tracking camera.

State-of-the-art imaging of neuromodulatory subcortical systems in aging and Alzheimer's disease: Challenges and opportunities

Primary prevention trials have shifted their focus to the earliest stages of Alzheimer's disease (AD). Autopsy data indicates that the neuromodulatory subcortical systems' (NSS) nuclei are specifically vulnerable to initial tau pathology, indicating that these nuclei hold great promise for early detection of AD in the context of the aging brain. The increasing availability of new imaging methods, ultra-high field scanners, new radioligands, and routine deep brain stimulation implants has led to a growing number of NSS neuroimaging studies on aging and neurodegeneration. Here, we review findings of current state-of-the-art imaging studies assessing the structure, function, and molecular changes of these nuclei during aging and AD. Furthermore, we identify the challenges associated with these imaging methods, important pathophysiologic gaps to fill for the AD NSS neuroimaging field, and provide future directions to improve our assessment, understanding, and clinical use of in vivo imaging of the NSS.

PET image enhancement using artificial intelligence for better characterization of epilepsy lesions

INTRODUCTION

[18F]fluorodeoxyglucose ([18F]FDG) brain PET is used clinically to detect small areas of decreased uptake associated with epileptogenic lesions, e.g., Focal Cortical Dysplasias (FCD) but its performance is limited due to spatial resolution and low contrast. We aimed to develop a deep learning-based PET image enhancement method using simulated PET to improve lesion visualization.

METHODS

We created 210 numerical brain phantoms (MRI segmented into 9 regions) and assigned 10 different plausible activity values (e.g., GM/WM ratios) resulting in 2100 ground truth high quality (GT-HQ) PET phantoms. With a validated Monte-Carlo PET simulator, we then created 2100 simulated standard quality (S-SQ) [18F]FDG scans. We trained a ResNet on 80% of this dataset (10% used for validation) to learn the mapping between S-SQ and GT-HQ PET, outputting a predicted HQ (P-HQ) PET. For the remaining 10%, we assessed Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), and Root Mean Squared Error (RMSE) against GT-HQ PET. For GM and WM, we computed recovery coefficients (RC) and coefficient of variation (COV). We also created lesioned GT-HQ phantoms, S-SQ PET and P-HQ PET with simulated small hypometabolic lesions characteristic of FCDs. We evaluated lesion detectability on S-SQ and P-HQ PET both visually and measuring the Relative Lesion Activity (RLA, measured activity in the reduced-activity ROI over the standard-activity ROI). Lastly, we applied our previously trained ResNet on 10 clinical epilepsy PETs to predict the corresponding HQ-PET and assessed image quality and confidence metrics.

RESULTS

Compared to S-SQ PET, P-HQ PET improved PNSR, SSIM and RMSE; significatively improved GM RCs (from 0.29 ± 0.03 to 0.79 ± 0.04) and WM RCs (from 0.49 ± 0.03 to 1 ± 0.05); mean COVs were not statistically different. Visual lesion detection improved from 38 to 75%, with average RLA decreasing from 0.83 ± 0.08 to 0.67 ± 0.14. Visual quality of P-HQ clinical PET improved as well as reader confidence.

CONCLUSION

P-HQ PET showed improved image quality compared to S-SQ PET across several objective quantitative metrics and increased detectability of simulated lesions. In addition, the model generalized to clinical data. Further evaluation is required to study generalization of our method and to assess clinical performance in larger cohorts.

Deep-TOF-PET: Deep learning-guided generation of time-of-flight from non-TOF brain PET images in the image and projection domains

We aim to synthesize brain time‐of‐flight (TOF) PET images/sinograms from their corresponding non‐TOF information in the image space (IS) and sinogram space (SS) to increase the signal‐to‐noise ratio (SNR) and contrast of abnormalities, and decrease the bias in tracer uptake quantification. One hundred forty clinical brain ¹⁸F‐FDG PET/CT scans were collected to generate TOF and non‐TOF sinograms. The TOF sinograms were split into seven time bins (0, ±1, ±2, ±3). The predicted TOF sinogram was reconstructed and the performance of both models (IS and SS) compared with reference TOF and non‐TOF. Wide‐ranging quantitative and statistical analysis metrics, including structural similarity index metric (SSIM), root mean square error (RMSE), as well as 28 radiomic features for 83 brain regions were extracted to evaluate the performance of the CycleGAN model. SSIM and RMSE of 0.99 ± 0.03, 0.98 ± 0.02 and 0.12 ± 0.09, 0.16 ± 0.04 were achieved for the generated TOF‐PET images in IS and SS, respectively. They were 0.97 ± 0.03 and 0.22 ± 0.12, respectively, for non‐TOF‐PET images. The Bland & Altman analysis revealed that the lowest tracer uptake value bias (−0.02%) and minimum variance (95% CI: −0.17%, +0.21%) were achieved for TOF‐PET images generated in IS. For malignant lesions, the contrast in the test dataset was enhanced from 3.22 ± 2.51 for non‐TOF to 3.34 ± 0.41 and 3.65 ± 3.10 for TOF PET in SS and IS, respectively. The implemented CycleGAN is capable of generating TOF from non‐TOF PET images to achieve better image quality.

Superresolution Reconstruction of Magnetic Resonance Images Based on a Nonlocal Graph Network

INTRODUCTION: High-resolution (HR) medical images are very important for doctors when diagnosing the internal pathological structures of patients and formulating precise treatment plans.OBJECTIVES: Other methods of superresolution cannot adequately capture nonlocal self-similarity information of images. To solve this problem, we proposed using graph convolution to capture non-local self-similar information.METHODS: This paper proposed a nonlocal graph network (NLGN) to perform single magnetic resonance (MR) image SR. Specifically, the proposed network comprises a nonlocal graph module (NLGM) and a nonlocal graph attention block (NLGAB). The NLGM is designed with densely connected residual blocks, which can fully explore the features of input images and prevent the loss of information. The NLGAB is presented to efficiently capture the dependency relationships among the given data by merging a nonlocal operation (NL) and a graph attention layer (GAL). In addition, to enable the current node to aggregate more beneficial information, when information is aggregated, we aggregate the neighbor nodes that are closest to the current node.RESULTS: For the scale r=2, the proposed NLGN achieves PSNR of 38.54 dB and SSIM of 0.9818 on the T(T1, BD) dataset, and yielding a 0.27 dB and 0.0008 improvement over the CSN method, respectively.CONCLUSION: The experimental results obtained on the IXI dataset show that the proposed NLGN performs better than the state-of-the-art methods.

Super-resolution acquisition and reconstruction for cone-beam SPECT with low-resolution detector

BACKGROUND AND OBJECTIVE

Single-photon emission computed tomography (SPECT) imaging, which provides information that reflects the human body's metabolic processes, has unique application value in disease diagnosis and efficacy evaluation. The imaging resolution of SPECT can be improved by exploiting high-performance detector hardware, but this exploitation generates high research and development costs. In addition, the inherent hardware structure of SPECT requires the use of a collimator, which limits the resolution in SPECT. The objective of this study is to propose a novel super-resolution (SR) reconstruction algorithm with two acquisition methods for cone-beam SPECT with low-resolution (LR) detector.

METHODS

A SR algorithm with two acquisition methods is proposed for cone-beam SPECT imaging in the projection domain. At each sampling angle, multi LR projections can be obtained by regularly moving the LR detector. For the two proposed acquisition methods, we develop a new SR reconstruction algorithm. Using our SR algorithm, a SR projection with the corresponding sampling angle can be obtained from multi LR projections via multi-iterations, and then, the SR SPECT image can be reconstructed. The peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), signal-to-noise ratio (SNR) and contrast recovery coefficient (CRC) are used to evaluate the final reconstruction quality.

RESULTS

The simulation results obtained under clean and noisy conditions verify the effectiveness of our SR algorithm. Three different phantoms are verified separately. 16 LR projections are obtained at each sampling angle, each with 32 × 32 bins. The high-resolution (HR) projection has 128 × 128 bins. The reconstruction result of the SR algorithm obtains an evaluation value that is almost the same as that of the HR reconstruction result. Our results indicate that the resolution of the resulting SPECT image is almost four times higher.

CONCLUSIONS

The authors develop a SR reconstruction algorithm with two acquisition methods for the cone-beam SPECT system. The simulation results obtained in clean and noisy environments prove that the SR algorithm has potential value, but it needs to be further tested on real equipment.

Deep learning-based image reconstruction and post-processing methods in positron emission tomography for low-dose imaging and resolution enhancement

Image processing plays a crucial role in maximising diagnostic quality of positron emission tomography (PET) images. Recently, deep learning methods developed across many fields have shown tremendous potential when applied to medical image enhancement, resulting in a rich and rapidly advancing literature surrounding this subject. This review encapsulates methods for integrating deep learning into PET image reconstruction and post-processing for low-dose imaging and resolution enhancement. A brief introduction to conventional image processing techniques in PET is firstly presented. We then review methods which integrate deep learning into the image reconstruction framework as either deep learning-based regularisation or as a fully data-driven mapping from measured signal to images. Deep learning-based post-processing methods for low-dose imaging, temporal resolution enhancement and spatial resolution enhancement are also reviewed. Finally, the challenges associated with applying deep learning to enhance PET images in the clinical setting are discussed and future research directions to address these challenges are presented.

Application of artificial intelligence in brain molecular imaging

Initial development of artificial Intelligence (AI) and machine learning (ML) dates back to the mid-twentieth century. A growing awareness of the potential for AI, as well as increases in computational resources, research, and investment are rapidly advancing AI applications to medical imaging and, specifically, brain molecular imaging. AI/ML can improve imaging operations and decision making, and potentially perform tasks that are not readily possible by physicians, such as predicting disease prognosis, and identifying latent relationships from multi-modal clinical information. The number of applications of image-based AI algorithms, such as convolutional neural network (CNN), is increasing rapidly. The applications for brain molecular imaging (MI) include image denoising, PET and PET/MRI attenuation correction, image segmentation and lesion detection, parametric image formation, and the detection/diagnosis of Alzheimer's disease and other brain disorders. When effectively used, AI will likely improve the quality of patient care, instead of replacing radiologists. A regulatory framework is being developed to facilitate AI adaptation for medical imaging.

Compensating Positron Range Effects of Ga-68 in Preclinical PET Imaging by Using Convolutional Neural Network: A Monte Carlo Simulation Study

This study aimed to investigate the feasibility of positron range correction based on three different convolutional neural network (CNN) models in preclinical PET imaging of Ga-68. The first model (CNN1) was originally designed for super-resolution recovery, while the second model (CNN2) and the third model (CNN3) were originally designed for pseudo CT synthesis from MRI. A preclinical PET scanner and 30 phantom configurations were modeled in Monte Carlo simulations, where each phantom configuration was simulated twice, once for Ga-68 (CNN input images) and once for back-to-back 511-keV gamma rays (CNN output images) with a 20 min emission scan duration. The Euclidean distance was used as the loss function to minimize the difference between CNN input and output images. According to our results, CNN3 outperformed CNN1 and CNN2 qualitatively and quantitatively. With regard to qualitative observation, it was found that boundaries in Ga-68 images became sharper after correction. As for quantitative analysis, the recovery coefficient (RC) and spill-over ratio (SOR) were increased after correction, while no substantial increase in coefficient of variation of RC (CV_RC) or coefficient of variation of SOR (CV_SOR) was observed. Overall, CNN3 should be a good candidate architecture for positron range correction in Ga-68 preclinical PET imaging.

Noise2Void: unsupervised denoising of PET images

Objective:Elevated noise levels in positron emission tomography (PET) images lower image quality and quantitative accuracy and are a confounding factor for clinical interpretation. The objective of this paper is to develop a PET image denoising technique based on unsupervised deep learning.Significance:Recent advances in deep learning have ushered in a wide array of novel denoising techniques, several of which have been successfully adapted for PET image reconstruction and post-processing. The bulk of the deep learning research so far has focused on supervised learning schemes, which, for the image denoising problem, require paired noisy and noiseless/low-noise images. This requirement tends to limit the utility of these methods for medical applications as paired training datasets are not always available. Furthermore, to achieve the best-case performance of these methods, it is essential that the datasets for training and subsequent real-world application have consistent image characteristics (e.g. noise, resolution, etc), which is rarely the case for clinical data. To circumvent these challenges, it is critical to develop unsupervised techniques that obviate the need for paired training data.Approach:In this paper, we have adapted Noise2Void, a technique that relies on corrupt images alone for model training, for PET image denoising and assessed its performance using PET neuroimaging data. Noise2Void is an unsupervised approach that uses a blind-spot network design. It requires only a single noisy image as its input, and, therefore, is well-suited for clinical settings. During the training phase, a single noisy PET image serves as both the input and the target. Here we present a modified version of Noise2Void based on a transfer learning paradigm that involves group-level pretraining followed by individual fine-tuning. Furthermore, we investigate the impact of incorporating an anatomical image as a second input to the network.Main Results:We validated our denoising technique using simulation data based on the BrainWeb digital phantom. We show that Noise2Void with pretraining and/or anatomical guidance leads to higher peak signal-to-noise ratios than traditional denoising schemes such as Gaussian filtering, anatomically guided non-local means filtering, and block-matching and 4D filtering. We used the Noise2Noise denoising technique as an additional benchmark. For clinical validation, we applied this method to human brain imaging datasets. The clinical findings were consistent with the simulation results confirming the translational value of Noise2Void as a denoising tool.

Artificial Intelligence-Based Image Enhancement in PET Imaging: Noise Reduction and Resolution Enhancement

High noise and low spatial resolution are two key confounding factors that limit the qualitative and quantitative accuracy of PET images. Artificial intelligence models for image denoising and deblurring are becoming increasingly popular for the postreconstruction enhancement of PET images. We present a detailed review of recent efforts for artificial intelligence-based PET image enhancement with a focus on network architectures, data types, loss functions, and evaluation metrics. We also highlight emerging areas in this field that are quickly gaining popularity, identify barriers to large-scale adoption of artificial intelligence models for PET image enhancement, and discuss future directions.

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.

A review of deep learning in medical imaging: Imaging traits, technology trends, case studies with progress highlights, and future promises

Since its renaissance, deep learning has been widely used in various medical imaging tasks and has achieved remarkable success in many medical imaging applications, thereby propelling us into the so-called artificial intelligence (AI) era. It is known that the success of AI is mostly attributed to the availability of big data with annotations for a single task and the advances in high performance computing. However, medical imaging presents unique challenges that confront deep learning approaches. In this survey paper, we first present traits of medical imaging, highlight both clinical needs and technical challenges in medical imaging, and describe how emerging trends in deep learning are addressing these issues. We cover the topics of network architecture, sparse and noisy labels, federating learning, interpretability, uncertainty quantification, etc. Then, we present several case studies that are commonly found in clinical practice, including digital pathology and chest, brain, cardiovascular, and abdominal imaging. Rather than presenting an exhaustive literature survey, we instead describe some prominent research highlights related to these case study applications. We conclude with a discussion and presentation of promising future directions.

Anatomy-guided PET reconstruction usingl1bowsher prior

Advances in simultaneous positron emission tomography/magnetic resonance imaging (PET/MRI) technology have led to an active investigation of the anatomy-guided regularized PET image reconstruction algorithm based on MR images. Among the various priors proposed for anatomy-guided regularized PET image reconstruction, Bowsher's method based on second-order smoothing priors sometimes suffers from over-smoothing of detailed structures. Therefore, in this study, we propose a Bowsher prior based on thel1-norm and an iteratively reweighting scheme to overcome the limitation of the original Bowsher method. In addition, we have derived a closed solution for iterative image reconstruction based on this non-smooth prior. A comparison study between the originall2and proposedl1Bowsher priors was conducted using computer simulation and real human data. In the simulation and real data application, small lesions with abnormal PET uptake were better detected by the proposedl1Bowsher prior methods than the original Bowsher prior. The originall2Bowsher leads to a decreased PET intensity in small lesions when there is no clear separation between the lesions and surrounding tissue in the anatomical prior. However, the proposedl1Bowsher prior methods showed better contrast between the tumors and surrounding tissues owing to the intrinsic edge-preserving property of the prior which is attributed to the sparseness induced byl1-norm, especially in the iterative reweighting scheme. Besides, the proposed methods demonstrated lower bias and less hyper-parameter dependency on PET intensity estimation in the regions with matched anatomical boundaries in PET and MRI. Therefore, these methods will be useful for improving the PET image quality based on the anatomical side information.

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.

Error propagation analysis of seven partial volume correction algorithms for [18F]THK-5351 brain PET imaging

BACKGROUND

Novel partial volume correction (PVC) algorithms have been validated by assuming ideal conditions of image processing; however, in real clinical PET studies, the input datasets include error sources which cause error propagation to the corrected outcome.

METHODS

We aimed to evaluate error propagations of seven PVCs algorithms for brain PET imaging with [¹⁸F]THK-5351 and to discuss the reliability of those algorithms for clinical applications. In order to mimic brain PET imaging of [¹⁸F]THK-5351, pseudo-observed SUVR images for one healthy adult and one adult with Alzheimer's disease were simulated from individual PET and MR images. The partial volume effect of pseudo-observed PET images were corrected by using Müller-Gärtner (MG), the geometric transfer matrix (GTM), Labbé (LABBE), regional voxel-based (RBV), iterative Yang (IY), structural functional synergy for resolution recovery (SFS-RR), and modified SFS-RR algorithms with incorporation of error sources in the datasets for PVC processing. Assumed error sources were mismatched FWHM, inaccurate image-registration, and incorrectly segmented anatomical volume. The degree of error propagations in ROI values was evaluated by percent differences (%diff) of PV-corrected SUVR against true SUVR.

RESULTS

Uncorrected SUVRs were underestimated against true SUVRs (- 15.7 and - 53.7% in hippocampus for HC and AD conditions), and application of each PVC algorithm reduced the %diff. Larger FWHM mismatch led to larger %diff of PVC-SUVRs against true SUVRs for all algorithms. Inaccurate image registration showed systematic propagation for most algorithms except for SFS-RR and modified SFS-RR. Incorrect segmentation of the anatomical volume only resulted in error propagations in limited local regions.

CONCLUSIONS

We demonstrated error propagation by numerical simulation of THK-PET imaging. Error propagations of 7 PVC algorithms for brain PET imaging with [¹⁸F]THK-5351 were significant. Robust algorithms for clinical applications must be carefully selected according to the study design of clinical PET data.

Machine learning in quantitative PET: A review of attenuation correction and low-count image reconstruction methods

The rapid expansion of machine learning is offering a new wave of opportunities for nuclear medicine. This paper reviews applications of machine learning for the study of attenuation correction (AC) and low-count image reconstruction in quantitative positron emission tomography (PET). Specifically, we present the developments of machine learning methodology, ranging from random forest and dictionary learning to the latest convolutional neural network-based architectures. For application in PET attenuation correction, two general strategies are reviewed: 1) generating synthetic CT from MR or non-AC PET for the purposes of PET AC, and 2) direct conversion from non-AC PET to AC PET. For low-count PET reconstruction, recent deep learning-based studies and the potential advantages over conventional machine learning-based methods are presented and discussed. In each application, the proposed methods, study designs and performance of published studies are listed and compared with a brief discussion. Finally, the overall contributions and remaining challenges are summarized.

PET image super-resolution using generative adversarial networks

The intrinsically low spatial resolution of positron emission tomography (PET) leads to image quality degradation and inaccurate image-based quantitation. Recently developed supervised super-resolution (SR) approaches are of great relevance to PET but require paired low- and high-resolution images for training, which are usually unavailable for clinical datasets. In this paper, we present a self-supervised SR (SSSR) technique for PET based on dual generative adversarial networks (GANs), which precludes the need for paired training data, ensuring wider applicability and adoptability. The SSSR network receives as inputs a low-resolution PET image, a high-resolution anatomical magnetic resonance (MR) image, spatial information (axial and radial coordinates), and a high-dimensional feature set extracted from an auxiliary CNN which is separately-trained in a supervised manner using paired simulation datasets. The network is trained using a loss function which includes two adversarial loss terms, a cycle consistency term, and a total variation penalty on the SR image. We validate the SSSR technique using a clinical neuroimaging dataset. We demonstrate that SSSR is promising in terms of image quality, peak signal-to-noise ratio, structural similarity index, contrast-to-noise ratio, and an additional no-reference metric developed specifically for SR image quality assessment. Comparisons with other SSSR variants suggest that its high performance is largely attributable to simulation guidance.