A review on AI in PET imaging

Deep transformer-based personalized dosimetry from SPECT/CT images: a hybrid approach for [177Lu]Lu-DOTATATE radiopharmaceutical therapy

PURPOSE

Accurate dosimetry is critical for ensuring the safety and efficacy of radiopharmaceutical therapies. In current clinical dosimetry practice, MIRD formalisms are widely employed. However, with the rapid advancement of deep learning (DL) algorithms, there has been an increasing interest in leveraging the calculation speed and automation capabilities for different tasks. We aimed to develop a hybrid transformer-based deep learning (DL) model that incorporates a multiple voxel S-value (MSV) approach for voxel-level dosimetry in [177Lu]Lu-DOTATATE therapy. The goal was to enhance the performance of the model to achieve accuracy levels closely aligned with Monte Carlo (MC) simulations, considered as the standard of reference. We extended our analysis to include MIRD formalisms (SSV and MSV), thereby conducting a comprehensive dosimetry study.

METHODS

We used a dataset consisting of 22 patients undergoing up to 4 cycles of [177Lu]Lu-DOTATATE therapy. MC simulations were used to generate reference absorbed dose maps. In addition, MIRD formalism approaches, namely, single S-value (SSV) and MSV techniques, were performed. A UNEt TRansformer (UNETR) DL architecture was trained using five-fold cross-validation to generate MC-based dose maps. Co-registered CT images were fed into the network as input, whereas the difference between MC and MSV (MC-MSV) was set as output. DL results are then integrated to MSV to revive the MC dose maps. Finally, the dose maps generated by MSV, SSV, and DL were quantitatively compared to the MC reference at both voxel level and organ level (organs at risk and lesions).

RESULTS

The DL approach showed slightly better performance (voxel relative absolute error (RAE) = 5.28 ± 1.32) compared to MSV (voxel RAE = 5.54 ± 1.4) and outperformed SSV (voxel RAE = 7.8 ± 3.02). Gamma analysis pass rates were 99.0 ± 1.2%, 98.8 ± 1.3%, and 98.7 ± 1.52% for DL, MSV, and SSV approaches, respectively. The computational time for MC was the highest (~2 days for a single-bed SPECT study) compared to MSV, SSV, and DL, whereas the DL-based approach outperformed the other approaches in terms of time efficiency (3 s for a single-bed SPECT). Organ-wise analysis showed absolute percent errors of 1.44 ± 3.05%, 1.18 ± 2.65%, and 1.15 ± 2.5% for SSV, MSV, and DL approaches, respectively, in lesion-absorbed doses.

CONCLUSION

A hybrid transformer-based deep learning model was developed for fast and accurate dose map generation, outperforming the MIRD approaches, specifically in heterogenous regions. The model achieved accuracy close to MC gold standard and has potential for clinical implementation for use on large-scale datasets.

The hidden world of protein aggregation

Though the book's journey into The Hidden World of Protein Aggregation has come to an end, the search for knowledge, the development of healthier lives, and the discovery of nature's mysteries continue, promising new horizons and discoveries yet to be discovered. The intricacies of protein misfolding and aggregation remain a mystery in cellular biology, despite advances made in unraveling them. In this chapter, we will summarize the specific conclusions from the previous chapters and explore the persistent obstacles and unanswered questions that motivate scientists to pursue exploration of protein misfolding and aggregation.

Stability of radiomic features from positron emission tomography images: a phantom study comparing advanced reconstruction algorithms and ordered subset expectation maximization

In this study, we compared the repeatability and reproducibility of radiomic features obtained from positron emission tomography (PET) images according to the reconstruction algorithm used-advanced reconstruction algorithms, such as HYPER iterative (IT), HYPER deep learning reconstruction (DLR), and HYPER deep progressive reconstruction (DPR), or traditional Ordered Subset Expectation Maximization (OSEM)-to understand the potential variations and implications of using advanced reconstruction techniques in PET-based radiomics. We used a heterogeneous phantom with acrylic spherical beads (4- or 8-mm diameter) filled with 18F. PET images were acquired and reconstructed using OSEM, IT, DLR, and DPR. Original and wavelet radiomic features were calculated using SlicerRadiomics. Radiomic feature repeatability was assessed using the Coefficient of Variance (COV) and intraclass correlation coefficient (ICC), and inter-acquisition time reproducibility was assessed using the concordance correlation coefficient (CCC). For the 4- and 8-mm diameter beads phantom, the proportion of radiomic features with a COV < 10% was equivocal or higher for the advanced reconstruction algorithm than for OSEM. ICC indicated that advanced methods generally outperformed OSEM in repeatability, except for the original features of the 8-mm beads phantom. In the inter-acquisition time reproducibility analysis, the combinations of 3 and 5 min exhibited the highest reproducibility in both phantoms, with IT and DPR showing the highest proportion of radiomic features with CCC > 0.8. Advanced reconstruction methods provided enhanced stability of radiomic features compared with OSEM, suggesting their potential for optimal image reconstruction in PET-based radiomics, offering potential benefits in clinical diagnostics and prognostics.

Deep learning-based PET image denoising and reconstruction: a review

This review focuses on positron emission tomography (PET) imaging algorithms and traces the evolution of PET image reconstruction methods. First, we provide an overview of conventional PET image reconstruction methods from filtered backprojection through to recent iterative PET image reconstruction algorithms, and then review deep learning methods for PET data up to the latest innovations within three main categories. The first category involves post-processing methods for PET image denoising. The second category comprises direct image reconstruction methods that learn mappings from sinograms to the reconstructed images in an end-to-end manner. The third category comprises iterative reconstruction methods that combine conventional iterative image reconstruction with neural-network enhancement. We discuss future perspectives on PET imaging and deep learning technology.

Applications of artificial intelligence in dementia

The recent evolution of artificial intelligence (AI) can be considered life-changing. In particular, there is great interest in emerging hot topics in AI such as image classification and natural language processing. Our world has been revolutionized by convolutional neural networks and transformer for image classification and natural language processing, respectively. Moreover, these techniques can be used in the field of dementia. We introduce some applications of AI systems for treating and diagnosing dementia, including image-classification AI for recognizing facial features associated with dementia, image-classification AI for classifying leukoaraiosis in MRI images, object-detection AI for detecting microbleeding in MRI images, object-detection AI for support care, natural language-processing AI for detecting dementia within conversations, and natural language-processing AI for chatbots. Such AI technologies can significantly transform the future of dementia diagnosis and treatment. Geriatr Gerontol Int 2024; 24: 25-30.

Artificial Intelligence's Transformative Role in Illuminating Brain Function in Long COVID Patients Using PET/FDG

Cutting-edge brain imaging techniques, particularly positron emission tomography with Fluorodeoxyglucose (PET/FDG), are being used in conjunction with Artificial Intelligence (AI) to shed light on the neurological symptoms associated with Long COVID. AI, particularly deep learning algorithms such as convolutional neural networks (CNN) and generative adversarial networks (GAN), plays a transformative role in analyzing PET scans, identifying subtle metabolic changes, and offering a more comprehensive understanding of Long COVID's impact on the brain. It aids in early detection of abnormal brain metabolism patterns, enabling personalized treatment plans. Moreover, AI assists in predicting the progression of neurological symptoms, refining patient care, and accelerating Long COVID research. It can uncover new insights, identify biomarkers, and streamline drug discovery. Additionally, the application of AI extends to non-invasive brain stimulation techniques, such as transcranial direct current stimulation (tDCS), which have shown promise in alleviating Long COVID symptoms. AI can optimize treatment protocols by analyzing neuroimaging data, predicting individual responses, and automating adjustments in real time. While the potential benefits are vast, ethical considerations and data privacy must be rigorously addressed. The synergy of AI and PET scans in Long COVID research offers hope in understanding and mitigating the complexities of this condition.

AI in Orthodontics: Revolutionizing Diagnostics and Treatment Planning-A Comprehensive Review

The advent of artificial intelligence (AI) in medicine has transformed various medical specialties, including orthodontics. AI has shown promising results in enhancing the accuracy of diagnoses, treatment planning, and predicting treatment outcomes. Its usage in orthodontic practices worldwide has increased with the availability of various AI applications and tools. This review explores the principles of AI, its applications in orthodontics, and its implementation in clinical practice. A comprehensive literature review was conducted, focusing on AI applications in dental diagnostics, cephalometric evaluation, skeletal age determination, temporomandibular joint (TMJ) evaluation, decision making, and patient telemonitoring. Due to study heterogeneity, no meta-analysis was possible. AI has demonstrated high efficacy in all these areas, but variations in performance and the need for manual supervision suggest caution in clinical settings. The complexity and unpredictability of AI algorithms call for cautious implementation and regular manual validation. Continuous AI learning, proper governance, and addressing privacy and ethical concerns are crucial for successful integration into orthodontic practice.

Reducing pediatric total-body PET/CT imaging scan time with multimodal artificial intelligence technology

OBJECTIVES

This study aims to decrease the scan time and enhance image quality in pediatric total-body PET imaging by utilizing multimodal artificial intelligence techniques.

METHODS

A total of 270 pediatric patients who underwent total-body PET/CT scans with a uEXPLORER at the Sun Yat-sen University Cancer Center were retrospectively enrolled. 18F-fluorodeoxyglucose (18F-FDG) was administered at a dose of 3.7 MBq/kg with an acquisition time of 600 s. Short-term scan PET images (acquired within 6, 15, 30, 60 and 150 s) were obtained by truncating the list-mode data. A three-dimensional (3D) neural network was developed with a residual network as the basic structure, fusing low-dose CT images as prior information, which were fed to the network at different scales. The short-term PET images and low-dose CT images were processed by the multimodal 3D network to generate full-length, high-dose PET images. The nonlocal means method and the same 3D network without the fused CT information were used as reference methods. The performance of the network model was evaluated by quantitative and qualitative analyses.

RESULTS

Multimodal artificial intelligence techniques can significantly improve PET image quality. When fused with prior CT information, the anatomical information of the images was enhanced, and 60 s of scan data produced images of quality comparable to that of the full-time data.

CONCLUSION

Multimodal artificial intelligence techniques can effectively improve the quality of pediatric total-body PET/CT images acquired using ultrashort scan times. This has the potential to decrease the use of sedation, enhance guardian confidence, and reduce the probability of motion artifacts.

Comparative Insight into Microglia/Macrophages-Associated Pathways in Glioblastoma and Alzheimer's Disease

Microglia and macrophages are pivotal to the brain's innate immune response and have garnered considerable attention in the context of glioblastoma (GBM) and Alzheimer's disease (AD) research. This review delineates the complex roles of these cells within the neuropathological landscape, focusing on a range of signaling pathways-namely, NF-κB, microRNAs (miRNAs), and TREM2-that regulate the behavior of tumor-associated macrophages (TAMs) in GBM and disease-associated microglia (DAMs) in AD. These pathways are critical to the processes of neuroinflammation, angiogenesis, and apoptosis, which are hallmarks of GBM and AD. We concentrate on the multifaceted regulation of TAMs by NF-κB signaling in GBM, the influence of TREM2 on DAMs' responses to amyloid-beta deposition, and the modulation of both TAMs and DAMs by GBM- and AD-related miRNAs. Incorporating recent advancements in molecular biology, immunology, and AI techniques, through a detailed exploration of these molecular mechanisms, we aim to shed light on their distinct and overlapping regulatory functions in GBM and AD. The review culminates with a discussion on how insights into NF-κB, miRNAs, and TREM2 signaling may inform novel therapeutic approaches targeting microglia and macrophages in these neurodegenerative and neoplastic conditions. This comparative analysis underscores the potential for new, targeted treatments, offering a roadmap for future research aimed at mitigating the progression of these complex diseases.

The OCDA-Net: a 3D convolutional neural network-based system for classification and staging of ovarian cancer patients using [18F]FDG PET/CT examinations

OBJECTIVE

To create the 3D convolutional neural network (CNN)-based system that can use whole-body [18F]FDG PET for recurrence/post-therapy surveillance in ovarian cancer (OC).

METHODS

In this study, 1224 image sets from OC patients who underwent whole-body [18F]FDG PET/CT at Kowsar Hospital between April 2019 and May 2022 were investigated. For recurrence/post-therapy surveillance, diagnostic classification as cancerous, and non-cancerous and staging as stage III, and stage IV were determined by pathological diagnosis and specialists' interpretation. New deep neural network algorithms, the OCDAc-Net, and the OCDAs-Net were developed for diagnostic classification and staging of OC patients using [18F]FDG PET/CT images. Examinations were divided into independent training (75%), validation (10%), and testing (15%) subsets.

RESULTS

This study included 37 women (mean age 56.3 years; age range 36-83 years). Data augmentation techniques were applied to the images in two phases. There were 1224 image sets for diagnostic classification and staging. For the test set, 170 image sets were considered for diagnostic classification and staging. The OCDAc-Net areas under the receiver operating characteristic curve (AUCs) and overall accuracy for diagnostic classification were 0.990 and 0.92, respectively. The OCDAs-Net achieved areas under the receiver operating characteristic curve (AUCs) of 0.995 and overall accuracy of 0.94 for staging.

CONCLUSIONS

The proposed 3D CNN-based models provide potential tools for recurrence/post-therapy surveillance in OC. The OCDAc-Net and the OCDAs-Net model provide a new prognostic analysis method that can utilize PET images without pathological findings for diagnostic classification and staging.

From FDG and beyond: the evolving potential of nuclear medicine

The radiopharmaceutical 2-[fluorine-18]fluoro-2-deoxy-D-glucose (FDG) has been dominantly used in positron emission tomography (PET) scans for over 20 years, and due to its vast utility its applications have expanded and are continuing to expand into oncology, neurology, cardiology, and infectious/inflammatory diseases. More recently, the addition of artificial intelligence (AI) has enhanced nuclear medicine diagnosis and imaging with FDG-PET, and new radiopharmaceuticals such as prostate-specific membrane antigen (PSMA) and fibroblast activation protein inhibitor (FAPI) have emerged. Nuclear medicine therapy using agents such as [177Lu]-dotatate surpasses conventional treatments in terms of efficacy and side effects. This article reviews recently established evidence of FDG and non-FDG drugs and anticipates the future trajectory of nuclear medicine.

[Evaluation of Low-dose Whole-body FDG PET with SiPM-based PET/CT Scanner: Visual and Semi-quantitative Analyses Using Random Sampling from Full-dose Scan Data]

PURPOSE

Sensitivity and count rate performance of the latest PET/CT scanners with a silicon photomultiplier (SiPM) have been substantially improved compared to scanners with a photomultiplier tube (PMT), thereby promising a low-dose whole-body PET scan with maintaining image quality. However, it is ethically difficult to verify the low-dose protocol in actual clinical settings. In this study, we investigated the effect of dose reduction on reconstructed images by using a low-dose simulation technique, i.e., reducing the number of events from the acquired data.

METHOD

For 21 subjects who underwent whole-body 18F-FDG PET examination with an SiPM-based PET/CT scanner, Biograph Vision (Siemens Healthineers, Erlangen, Germany), at a dosage of 3.5 MBq/kg and a continuous bed motion speed of 1.1 mm/sec (the standard protocol in our hospital), the number of events in acquired list data (100%; "full-dose") was reduced to 50%, 25%, 12.5%, and 6.25% ("low-dose"). The low-dose reconstructed images were evaluated visually and physically with reference to the full-dose images. The physical evaluation was performed by calculating differences in SUVmax at abnormal uptake (n=54) between the full-dose and low-dose images.

RESULT

The 25% data images were visually acceptable, and the difference in SUVmax between the 100% and 25% data images was 9.8±13.5%.

CONCLUSION

Our results suggest that Biograph Vision is a feasible method to reduce conventional dose with the potential use of 25% data images.

Quantitation of dynamic total-body PET imaging: recent developments and future perspectives

BACKGROUND

Positron emission tomography (PET) scanning is an important diagnostic imaging technique used in disease diagnosis, therapy planning, treatment monitoring, and medical research. The standardized uptake value (SUV) obtained at a single time frame has been widely employed in clinical practice. Well beyond this simple static measure, more detailed metabolic information can be recovered from dynamic PET scans, followed by the recovery of arterial input function and application of appropriate tracer kinetic models. Many efforts have been devoted to the development of quantitative techniques over the last couple of decades.

CHALLENGES

The advent of new-generation total-body PET scanners characterized by ultra-high sensitivity and long axial field of view, i.e., uEXPLORER (United Imaging Healthcare), PennPET Explorer (University of Pennsylvania), and Biograph Vision Quadra (Siemens Healthineers), further stimulates valuable inspiration to derive kinetics for multiple organs simultaneously. But some emerging issues also need to be addressed, e.g., the large-scale data size and organ-specific physiology. The direct implementation of classical methods for total-body PET imaging without proper validation may lead to less accurate results.

CONCLUSIONS

In this contribution, the published dynamic total-body PET datasets are outlined, and several challenges/opportunities for quantitation of such types of studies are presented. An overview of the basic equation, calculation of input function (based on blood sampling, image, population or mathematical model), and kinetic analysis encompassing parametric (compartmental model, graphical plot and spectral analysis) and non-parametric (B-spline and piece-wise basis elements) approaches is provided. The discussion mainly focuses on the feasibilities, recent developments, and future perspectives of these methodologies for a diverse-tissue environment.

Pretreatment 18F-FDG uptake heterogeneity may predict treatment outcome of combined Trastuzumab and Pertuzumab therapy in patients with metastatic HER2 positive breast cancer

OBJECTIVE

Intra-tumoral heterogeneity of 18F-fluorodeoxyglucose (18F-FDG) uptake has been proven to be a surrogate marker for predicting treatment outcome in various tumors. However, the value of intra-tumoral heterogeneity in metastatic Human epidermal growth factor receptor 2(HER2) positive breast cancer (MHBC) remains unknown. The aim of this study was to evaluate 18F-FDG uptake heterogeneity to predict the treatment outcome of the dual target therapy with Trastuzumab and Pertuzumab(TP) in MHBC.

METHODS

Thirty-two patients with MHBC who underwent 18F-FDG positron emission tomography/computed tomography (PET/CT) scan before TP were enrolled retrospectively. The region of interesting (ROI) of the lesions were drawn, and maximum standard uptake value (SUVmax), mean standard uptake value (SUVmean), total lesion glycolysis (TLG), metabolic tumor volume (MTV) and heterogeneity index (HI) were recorded. Correlation between PET/CT parameters and the treatment outcome was analyzed by Spearman Rank Test. The ability to predict prognosis were determined by time-dependent survival receiver operating characteristic (ROC) analysis. And the survival analyses were then estimated by Kaplan-Meier method and compared by log-rank test.

RESULTS

The survival analysis showed that HI50% calculated by delineating the lesion with 50%SUVmax as threshold was a significant predictor of patients with MHBC treated by the treatment with TP. Patients with HI50% (≥ 1.571) had a significantly worse prognosis of progression free survival (PFS) (6.87 vs. Not Reach, p = 0.001). The area under curve (AUC), the sensitivity and the specificity were 0.88, 100% and 63.6% for PFS, respectively.

CONCLUSION

18F-FDG uptake heterogeneity may be useful for predicting the prognosis of MHBC patients treated by TP.

[Opportunities for artificial intelligence in radiation protection : Improving safety of diagnostic imaging]

CLINICAL/METHODOLOGICAL ISSUE

Imaging of structures of internal organs often requires ionizing radiation, which is a health risk. Reducing the radiation dose can increase the image noise, which means that images provide less information.

STANDARD RADIOLOGICAL METHODS

This problem is observed in commonly used medical imaging modalities such as computed tomography (CT), positron emission tomography (PET), single photon emission computed tomography (SPECT), angiography, fluoroscopy, and any modality that uses ionizing radiation for imaging.

METHODOLOGICAL INNOVATIONS

Artificial intelligence (AI) can improve the quality of low-dose images and help minimize radiation exposure. Potential applications are explored, and frameworks and procedures are critically evaluated.

PERFORMANCE

The performance of AI models varies. High-performance models could be used in clinical settings in the near future. Several challenges (e.g., quantitative accuracy, insufficient training data) must be addressed for optimal performance and widespread adoption of this technology in the field of medical imaging.

PRACTICAL RECOMMENDATIONS

To fully realize the potential of AI and deep learning (DL) in medical imaging, research and development must be intensified. In particular, quality control of AI models must be ensured, and training and testing data must be uncorrelated and quality assured. With sufficient scientific validation and rigorous quality management, AI could contribute to the safe use of low-dose techniques in medical imaging.

Generative adversarial network-created brain SPECTs of cerebral ischemia are indistinguishable to scans from real patients

Deep convolutional generative adversarial networks (GAN) allow for creating images from existing databases. We applied a modified light-weight GAN (FastGAN) algorithm to cerebral blood flow SPECTs and aimed to evaluate whether this technology can generate created images close to real patients. Investigating three anatomical levels (cerebellum, CER; basal ganglia, BG; cortex, COR), 551 normal (248 CER, 174 BG, 129 COR) and 387 pathological brain SPECTs using N-isopropyl p-I-123-iodoamphetamine (¹²³I-IMP) were included. For the latter scans, cerebral ischemic disease comprised 291 uni- (66 CER, 116 BG, 109 COR) and 96 bilateral defect patterns (44 BG, 52 COR). Our model was trained using a three-compartment anatomical input (dataset 'A'; including CER, BG, and COR), while for dataset 'B', only one anatomical region (COR) was included. Quantitative analyses provided mean counts (MC) and left/right (LR) hemisphere ratios, which were then compared to quantification from real images. For MC, 'B' was significantly different for normal and bilateral defect patterns (P < 0.0001, respectively), but not for unilateral ischemia (P = 0.77). Comparable results were recorded for LR, as normal and ischemia scans were significantly different relative to images acquired from real patients (P ≤ 0.01, respectively). Images provided by 'A', however, revealed comparable quantitative results when compared to real images, including normal (P = 0.8) and pathological scans (unilateral, P = 0.99; bilateral, P = 0.68) for MC. For LR, only uni- (P = 0.03), but not normal or bilateral defect scans (P ≥ 0.08) reached significance relative to images of real patients. With a minimum of only three anatomical compartments serving as stimuli, created cerebral SPECTs are indistinguishable to images from real patients. The applied FastGAN algorithm may allow to provide sufficient scan numbers in various clinical scenarios, e.g., for "data-hungry" deep learning technologies or in the context of orphan diseases.

Automatic detection of primary and metastatic lesions on cervicothoracic region and whole-body bone using a uniform machine-learnable approach for [18F]-FDG-PET/CT image analysis

We propose a method to detect primary and metastatic lesions with Fluorine−18 fluorodeoxyglucose (FDG) accumulation in the lung field, neck, mediastinum, and bony regions on the FDG-PET/CT images. To search for systemic lesions, various anatomical structures must be considered. The proposed method is addressed by using an extraction process for anatomical regions and a uniform lesion detection approach. The uniform approach does not utilize processes that reflect any region-specific anatomical aspects but has a machine-learnable framework. Therefore, it can work as a lesion detection process for a specific anatomical region if it machine-learns the specific region data. In this study, three lesion detection processes for the whole-body bone region, lung field, or neck-mediastinum region are obtained. These detection processes include lesion candidate detection and false positive (FP) candidate elimination. The lesion candidate detection is based on a voxel anomaly detection with a one-class support vector machine. The FP candidate elimination is performed using an AdaBoost classifier ensemble. The image features used by the ensemble are selected sequentially during training and are optimal for candidate classification. Three-fold cross-validation was used to detect performance with the 54 diseased FDG-PET/CT images. The mean sensitivity for detecting primary and metastatic lesions at 3 FPs per case was 0.89 with a 0.10 standard deviation (SD) in the bone region, 0.80 with a 0.10 SD in the lung field, and 0.87 with a 0.10 SD in the neck region. The average areas under the ROC curve were 0.887 with a 0.125 SD for detecting bone metastases, 0.900 with a 0.063 SD for detecting pulmonary lesions, and 0.927 with a 0.035 SD for detecting the neck-mediastinum lesions. These detection performances indicate that the proposed method could be applied clinically. These results also show that the uniform approach has high versatility for providing various lesion detection processes.

3D Convolutional Neural Network-Based Denoising of Low-Count Whole-Body 18F-Fluorodeoxyglucose and 89Zr-Rituximab PET Scans

Acquisition time and injected activity of ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) PET should ideally be reduced. However, this decreases the signal-to-noise ratio (SNR), which impairs the diagnostic value of these PET scans. In addition, ⁸⁹Zr-antibody PET is known to have a low SNR. To improve the diagnostic value of these scans, a Convolutional Neural Network (CNN) denoising method is proposed. The aim of this study was therefore to develop CNNs to increase SNR for low-count ¹⁸F-FDG and ⁸⁹Zr-antibody PET. Super-low-count, low-count and full-count ¹⁸F-FDG PET scans from 60 primary lung cancer patients and full-count ⁸⁹Zr-rituximab PET scans from five patients with non-Hodgkin lymphoma were acquired. CNNs were built to capture the features and to denoise the PET scans. Additionally, Gaussian smoothing (GS) and Bilateral filtering (BF) were evaluated. The performance of the denoising approaches was assessed based on the tumour recovery coefficient (TRC), coefficient of variance (COV; level of noise), and a qualitative assessment by two nuclear medicine physicians. The CNNs had a higher TRC and comparable or lower COV to GS and BF and was also the preferred method of the two observers for both ¹⁸F-FDG and ⁸⁹Zr-rituximab PET. The CNNs improved the SNR of low-count ¹⁸F-FDG and ⁸⁹Zr-rituximab PET, with almost similar or better clinical performance than the full-count PET, respectively. Additionally, the CNNs showed better performance than GS and BF.