Segmentation of Head and Neck Tumors Using Dual PET/CT Imaging:Comparative Analysis of 2D,2.5D,and 3D Approaches Using UNet Transformer

The segmentation of head and neck(H&N)tumors in dual Positron Emission Tomography/Computed Tomogra-phy(PET/CT)imaging is a critical task in medical imaging,providing essential information for diagnosis,treatment planning,and outcome prediction.Motivated by the need for more accurate and robust segmentation methods,this study addresses key research gaps in the application of deep learning techniques to multimodal medical images.Specifically,it investigates the limitations of existing 2D and 3D models in capturing complex tumor structures and proposes an innovative 2.5D UNet Transformer model as a solution.The primary research questions guiding this study are:(1)How can the integration of convolutional neural networks(CNNs)and transformer networks enhance segmentation accuracy in dual PET/CT imaging?(2)What are the comparative advantages of 2D,2.5D,and 3D model configurations in this context?To answer these questions,we aimed to develop and evaluate advanced deep-learning models that leverage the strengths of both CNNs and transformers.Our proposed methodology involved a comprehensive preprocessing pipeline,including normalization,contrast enhancement,and resampling,followed by segmentation using 2D,2.5D,and 3D UNet Transformer models.The models were trained and tested on three diverse datasets:HeckTor2022,AutoPET2023,and SegRap2023.Performance was assessed using metrics such as Dice Similarity Coefficient,Jaccard Index,Average Surface Distance(ASD),and Relative Absolute Volume Difference(RAVD).The findings demonstrate that the 2.5D UNet Transformer model consistently outperformed the 2D and 3D models across most metrics,achieving the highest Dice and Jaccard values,indicating superior segmentation accuracy.For instance,on the HeckTor2022 dataset,the 2.5D model achieved a Dice score of 81.777 and a Jaccard index of 0.705,surpassing other model configurations.The 3D model showed strong boundary delineation performance but exhibited variability across datasets,while the 2D model,although effective,generally underperformed compared to its 2.5D and 3D counterparts.Compared to related literature,our study confirms the advantages of incorporating additional spatial context,as seen in the improved performance of the 2.5D model.This research fills a significant gap by providing a detailed comparative analysis of different model dimensions and their impact on H&N segmentation accuracy in dual PET/CT imaging.

In Vivo Tumor-Targeted Dual-Modality PET/Optical Imaging with a Yolk/Shell-Structured Silica Nanosystem

Silica nanoparticles have been one of the most promising nanosystems for biomedical applications due to their facile surface chemistry and non-toxic nature. However, it is still challenging to effectively deliver them into tumor sites and noninvasively visualize their in vivo biodistribution with excellent sensitivity and accuracy for effective cancer diagnosis. In this study, we design a yolk/shell-structured silica nanosystem ^(64) Cu-NOTAQD@HMSN-PEG-TRC105, which can be employed for tumor vasculature targeting and dual-modality PET/optical imaging, leading to superior targeting specificity, excellentimaging capability and more reliable diagnostic outcomes.By combining vasculature targeting, pH-sensitive drug delivery, and dual-modality imaging into a single platform,as-designed yolk/shell-structured silica nanosystems may be employed for the future image-guided tumor-targeted drug delivery, to further enable cancer theranostics.

Microglia activation in the offspring of prenatal Poly I:C exposed rats:a PET imaging and immunohistochemistry study

Background The well-known ‘pyrotherapy’ of Julius Wagner-Jauregg might be the beginning of the study on the immunological concepts of schizophrenia. As the primary immune effector cells in the brain, microglia play a pivotal role in neuroinfammatory processes. Maternal viral infection during pregnancy is associated with an increased risk for psychiatric disorders with presumed neurodevelopmental origin, including autism spectrum disorders and schizophrenia. The present study was to quantify microglia activation in vivo in the mature offspring of rats exposed to polyriboinosinic–polyribocytidilicacid （Poly I：C） during pregnancy using ^11C-PK11195 positron emission tomography （PET） and immunohistochemistry.Objective The study aimed to quantify microglia activation in vivo in the prefrontal cortex and hippocampus in mature offspring of prenatal Poly I：C exposed rats.Methods Offspring of Poly I：C-treated dams were the model group, offspring of saline-treated dams were the control group. Behavioural test for two groups was taken by spontaneous activity, prepulse inhibition （PPI） and latent inhibition （LI） test （including active avoidance conditioning task and passive avoidance conditioning task）. Randomly selected successful model rats were assessed by behavioural test in the model group and control group rats. 11C-PK11195 micro-PET/CT and immunohistochemistry were performed on the selected rats to measure microglia activation.Results The treatment group showed hyperlocomotion and defcits in PPI and LI compared with the control group. The treatment group also showed an increased 11C-PK11195 uptake ratio in the prefrontal cortex （t=-3.990, p=0.003） and hippocampus （t=-4.462, p=0.001）. The number of activated microglia cells was signifcantly higher in the treatment group than in the control group （hippocampus： t=8.204, p〈0.001; prefrontal： t=6.995, p〈0.001）. Within the treatment group, there were signifcant correlations between the behavioural parameters and the activation of microglia as measured by PET and immunohistochemistry.Conclusions The present study demonstrated microglia activation in vivo in the prefrontal cortex and hippocampus in mature offspring of prenatal Poly I：C exposed rats. This study suggests that microglia activation may play a possible or potential role in the pathogenesis of schizophrenia.

The metabolic brain network in patients with Parkinson's disease based on ^(18)F-FDG PET imaging: evaluation of neuronal injury and regeneration

Over the past two decades, the development of functional imaging methods has greatly promoted our understanding on the changes of neurons following neurodegenerative disorders, such as Parkin- son＇s disease （PD）. The application of a spatial covariance analysis on 18F-FDG PET imaging has led to the identification of a distinc- tive disease-related metabolic pattern. This pattern has proven to be useful in clinical diagnosis, disease progression monitoring as well as assessment of the neuronal changes before and after clinical treatment. It may potentially serve as an objective biomarker on disease progression monitoring, assessment, histological and func- tional evaluation of related diseases.

Synthesis and biological evaluation of ^(18)F-FB-NGA as a hepatic asialoglycoprotein receptor PET imaging agent

Asialoglycoprotein receptor(ASGP-R)is a hepatic membrane receptor that uniquely exists on the surface of mammalian hepatocytes,and has been used as target of liver functional imaging agents for many years.We labeled the Galactosyl-neoglycoalbumin(NGA)with 18F to get a PET molecular probe 18F-FB-NGA and evaluated its ability as a liver functional PET imaging agent.The 18F-FB-NGA was prepared with NGA by conjugation with Nsuccinimidyl-4-18F-fluorobenzoate(18F-SFB)and purified with PD-10 desalting column.The radiolabeling yield and radiochemical purity of 18F-FB-NGA were determined by radio-HPLC.Starting with 18F-F–,the total time for 18F-FB-NGA was about 120±10 min.The decay-corrected radiochemical yield is about 25–30%.The radiochemical purity of purified 18F-FB-NGA was more than 98%.Labeled with 185–1850 MBq 18F-SFB,the specific activity of 18F-FBNGA was estimated to be 7.83–78.3 TBq/mmol.Biodistribution of 18F-FB-NGA in normal mice was investigated after injection through the tail vein.The results showed that the liver accumulated 39.47±3.42 and 12.12±6.11%ID/g at 10 and 30 min after injection,respectively.Dynamic MicroPET images in mice were acquired with and without block after injection of the radiotracer,respectively.High liver activity accumulation was observed at 5 min after injection in normal group.On the contrary,the liver accumulation was significantly lower after block,indicating the specific binding to ASGP-R.18F-FB-NGA is probably a potential PET liver imaging agent.

Synthesis and evaluation of ^(64)Cu-radiolabeled NOTA-cetuximab(^(64)Cu-NOTA-C225) for immuno-PET imaging of EGFR expression

Objective: Epidermal growth factor receptor(EGFR) is overexpressed in a wide variety of solid tumors, serving as a well-characterized target for cancer imaging or therapy. In this study, we aimed to design and synthesize a radiotracer, ^(64) Cu-NOTA-C225, targeting EGFR for tumor positron emission tomography(PET) imaging.Methods: Cetuximab(C225) was conjugated to a bifunctional chelator, p-isothiocyanatobenzyl-1,4,7-triazacyclononane-1,4,7-triacetic acid(NOTA), and further radiolabeled with copper-^(64) for PET imaging. ^(64) CuNOTA-IgG and Cy5.5-C225 were also synthesized as control probes. A431 and A549 mouse models were established for micro-PET and/or near-infrared fluorescence(NIRF) imaging.Results: ^(64) Cu-NOTA-C225 exhibited stability in vivo and in vitro up to 24 h and 50 h post-injection,respectively. A431 tumors with average standard uptake values(SUVs) of 5.61±0.69, 6.68±1.14, 7.80±1.51 at 6, 18 and 36 h post-injection, respectively, which were significantly higher than that of moderate EGFR expressing tumors(A549), with SUVs of 0.89±0.16, 4.70±0.81, 2.01±0.50 at 6, 18 and 36 h post-injection, respectively. The expression levels of A431 and A549 were confirmed by western blotting. Additionally, the tracer uptake in A431 tumors can be blocked by unlabeled cetuximab, suggesting that tracer uptake by tumors was receptor-mediated.Furthermore, NIRF imaging using Cy5.5-C225 showed that the fluorescence intensity in tumors increased with time, with a maximal intensity of 8.17 E+10(p/s/cm^2/sr)/(μW/cm^2) at 48 h post-injection, which is consistent with the paradigm from micro-PET imaging in A431 tumor-bearing mice.Conclusions: The ^(64) Cu-NOTA-C225 PET imaging may be able to specifically and sensitively differentiate tumor models with different EGFR expression levels. It offers potentials as a PET radiotracer for imaging of tracer EGFR-positive tumors.

AN AUTOMATIC SEGMENTATION METHOD FOR FAST IMAGING IN PET

A new segmentation method has been developed for PET fast imaging. The technique automatically segments the transmission images into different anatomical regions, it efficiently reduced the PET transmission scan time. The result shows that this method gives only 3 min-scan time which is perfect for attenuation correction of the PET images instead of the original 15-30 min-scan time. This approach has been successfully tested both on phantom and clinical data.

Imaging performance evaluation in depth-of-interaction PET with a new method of sinogram generation:A Monte Carlo simulation study

In conventional PET systems,the parallax error degrades image resolution and causes image distortion.To remedy this,a PET ring diameter has to be much larger than the required size of field of view(FOV),and therefore the cost goes up.Measurement of depth-of-interaction(DOI)information is effective to reduce the parallax error and improve the image quality.This study is aimed at developing a practical method to incorporate DOI information in PET sinogram generation and image reconstruction processes and evaluate its efficacy through Monte Carlo simulation.An animal PET system with 30-mm long LSO crystals and 2-mm DOI measurement accuracy was simulated and list-mode PET data were collected.A sinogram generation method was proposed to bin each coincidence event to the correct LOR location according to both incident crystal indices and DOI positions of the two annihilation photons.The sinograms were reconstructed with an iterative OSMAPEM(ordered subset maximum a posteriori expectation maximization)algorithm.Two phantoms(a rod source phantom and a Derenzo phantom)were simulated,and the benefits of DOI were investigated in terms of reconstructed source diameter(FWHM)and source positioning accuracy.The results demonstrate that the proposed method works well to incorporate DOI information in data processing,which not only overcomes the image distortion problem but also significantly improves image resolution and resolution uniformity and results in satisfactory image quality.

Glutamine Metabolism: A New Target Pathway for PET Imaging and Cancer Therapy

Individuals with metabolic syndrome will have increased risk for cardiovascular disease, diabetes and cancer. Based on abnormal glucose metabolism, or called Warburg effect in cancer cells, positron emission tomography (PET) imaging using 18F-FDG has achieved a great success in diagnosis and monitoring therapeutic responds for cancer patients. Beyond glucose metabolism, resurgence of glutamine metabolism in cancer research has recently broadened interests. Observations of addiction to glutamine in cancer cells lead to considering contribution of glutaminolysis in cancer cell growth, differentiation and proliferation. Furthermore, oncogenes have been found to be major factors in modulating abnormal glutamine metabolism in cancer cells. PET imaging probes and therapeutic agents targeting glutamine metabolic and signaling pathways have been proposed and investigated.

Hypoxia Imaging of Rodent Xenografts with ¹⁸F-Fluoromisonidazole: Comparison of Dynamic and Static PET Imaging

Purpose: To generate parametric images of tumor hypoxia in a tumor-bearing rat model using voxel-based compartmental analysis of dynamic fluorine-18 labeled misonidazole (18F-FMISO) microPET? images, and to compare the parametric images thus derived with static “late” 18F-FMISO microPET? images for the detection of tumor hypoxia. Materials and Methods: Nude rats bearing HT-29 colorectal carcinoma xenografts (≈1.5 - 2 cm in diameter) in the right hind limb were positioned in a custom-fabricated, animal-specific foam mold. Animals were injected via the tail vein with ≈55.5 MBq 18F-FMISO and continuously imaged for either 60 or 120 minutes, with additional late static images up to 3 hour post-injection. The raw list-mode data was reconstructed into 37 - 64 frames with earlier frames of shorter time durations (12 - 15 seconds) and later frames of longer durations (up to 300 seconds). Time activity curves (TACs) were generated over regions encompassing the tumor as well as an artery, the latter for use as an input function. A beta version of a compartmental modeling package (BioGuide?, Philips Healthcare) was used to generate parametric images of k3 and Ki, rate constants of entrapment and flux of 18F-FMISO, respectively. Results: Data for 7 HT-29 tumor xenografts were presented, 6 of which yielded clear areas of tumor hypoxia as defined by Ki/k3 maps. Importantly, intratumoral foci with high 18F-FMISO uptakes on the late images did not always exhibit high Ki/k3 values and may there- fore represent false-positives for radiobiologically significant hypoxia. Conclusions: This study attempts to quantify tumor hypoxia using compartmental analysis of dynamic 18F-FMISO PET images in rodent xenograft tumor models. The results demonstrate feasibility of the approach in small-animal imaging studies, and provide evidence for the possible unreliability of late-time static imaging of 18F-FMISO PET in identifying tumor hypoxia.

Application of ^(18) F-FDG PET/CT Imaging in Diagnosing Bladder Tumor Metastasis Lesions

Bladder tumor is the most common malignant tumor in urinary system and always com- panied with lymph node metastasis. The accurate staging plays a significant role in treatment for bladder tumor and prognostic evaluation, and the distant metastasis predicts worse prognosis. The objective of this study was to assess the clinical significance of 18F-FDG PET/CT imaging in diagnosing bladder tumor metastasis lesions. A retrospective analysis of 60 patients with bladder tumor from October 2008 to May 2010 was done. The patients were stratified based on the imaging technique. Among all 60 cases, besides the primary lesion, 81 suspected lesions were spotted and 73 confirmed as metastasis, including 50 lymph node metastases, 22 distant metastases, and 1 bone metastasis. For PET/CT imaging, its sensitivity was 94.5%, specificity 87.5%, positive predictive value 98.6%, negative predictive value 63.6% and accuracy 93.8% respectively. For CT, its sensitivity was 82.2%, specificity 50%, positive predictive value 93.8%, negative predictive value 23.5% and accuracy 79% respectively. PET/CT im- aging was superior to CT in sensitivity, specificity and accuracy. In conclusion, 18F-FDG PET/CT imaging is more significant in diagnosing bladder tumor metastasis lesions.

Design of a Novel Front-End Readout ASIC for PET Imaging System

The architecture of a multi-channel front-end system is important for realizing a high-resolution PET system. We propose a novel front-end readout electronic system with TDC to deal with time information for PET system which can easily design the timing control. Each channel consists of a charge preamplifier, slow/fast shaper, discriminator and an analog memory. There are an ADC and a TDC to process the energy information and time information for each channel at the same time. In this paper, the whole system signals flow is all simulated by MATLAB. The simulation results show that the proposed system can process slender current from the detector and achieve the energy and time information. The proposed architecture can be applied to high-resolution PET imaging systems with multi-channel ASICs.

Strategies to Improve CT Dose Optimization for Hybrid PET/CT Imaging

Background: To evaluate whether current dose reduction strategies for the CT component of hybrid Positron Emission Tomography—Computed Tomography (PET/CT) systems could reduce patient dose with maintaining adequate image quality for PET/CT studies. Materials and Methods: Literature survey was initially based on the selection of keywords and years of publication to identify potentially relevant articles, then the further search was conducted on the authors and references from these articles. The abstract of each article was first appraised to decide whether the content was relevant to this research question. The articles were classified into five groups: studies on dosimetry, studies on radiation-induced diseases, studies on dose reduction methods for CT-based attenuation correction (CTAC), studies on dose reduction methods for CT localization, and studies on reducing the need for a full-dose diagnostic CT in PET/CT imaging. 58 peer-reviewed articles were selected and appraised and 29 articles were used to compose this literature review. Results: The published nuclear medicine and medical physics literature were reviewed. CT dose contributed 47% - 81% of the total effective dose of a standard PET/CT study and was associated with radiation-induced diseases. The dose reduction techniques were extracted and divided into three categories: reducing the CT dose for attenuation correction (AC) and localization, selectively localizing CT use, and reducing the need for a full-dose diagnostic CT. Conclusion: Three strategies have been demonstrated, with high potential for reducing patient dose while maintaining an adequate CT image quality, used for CTCA localization and diagnosis, respectively.

Synthesis and Radiation Dosimetry of [⁶⁸Ga]-Ga-Lys¹, Lys³-DOTA-Bombesin (1,14) Antagonist for PET-Imaging, as a Potential Theragnostic Tracer in Oncology

This Bombesin (BBN), a tetradecapeptide analog of human gastrin-releasing peptide (GRP) with a high binding affinity for GRP receptors (GRPR), is over- expressed in early stages of androgen-dependent prostate carcinomas, but not in advanced stages. Therefore, there is a need to develop effective tracers for the accurate and specific detection of this disease. The objective of this study was to evaluate Lys1, Lys3-DOTA-BBN (1,14) analog with the radiolabeled positron emitter [68Ga]-Ga-BBN for receptor imaging with PET, and to determine its biodistribution and radiation dosimetry using whole-body (WB) PET scans in healthy volunteers. The highest uptake was in the pancreas, followed by urinary bladder. The critical organ was pancreas with a mean absorbed dose of 206 ± 0.7, 210 ± 0.7, 120 ± 0.9, 390.23 ± 0.6 μGy/MBq and the effective doses were estimated as 73.2 ± 0.6, 49.8 ± 0.3 μGy/MBq (women and men, respectively).

^(18)F-FETNIM PET/CT hypoxia imaging in non-small cell lung cancer:preliminary clinical observation

Objective:The aims of this study were to evaluate potential side effects of 18F-fluoroerythronitroimidazole (18F-FETNIM) as a new-type hypoxia-imaging agent and to investigate the feasibility of 18F-FETNIM PET imaging in advanced non-small cell lung cancer (NSCLC) patients and the correlations of hypoxia extent with tumor volume or pathological type. Methods: Twenty-six NSCLC patients were prospectively included in the study. PET/CT scans were performed 2 h after intravenous injection of 18F-FETNIM in all 26 patients. A pixel-by-pixel calculation of tumor to blood (T/B) activity ratio for all image planes was calculated. The number of pixels in the tumor volume with a T/B ratio≥ 1.5,indicating significant hypoxia,was determined and converted to mL units to measure the hypoxia volume (HV). Results: The images were clearly identified after 2 h post-injection of 18F-FETNIM. The tumors in 4 cases were not distinguished from background,while the remaining 22 displayed local 18F-FETNIM uptake in thoracic lesions moderately to markedly higher than background. There was no correlation between 18F-FETNIM uptake with pathological type. There were significant correlations of HV and also the T/B ratio with tumor volume. Conclusion:18F-FETNIM is a promising hypoxia-imaging agent which clinical use is safe and satisfactory. The preliminary study provides valuable methods and experience to its further research.

PET imaging of cerebral metabolic change in tinnitususing ^(18)F-FDG

Tinnitus is an auditory disorder hardly assessable by clinical technology. PET imaging of the brain in 13 cases with and 10 without tinnitus was undertaken at 40 min after injection of 280-440 MBq 18F-FDG. To ensure the quality of the PET study, all cases followed a normalized procedure with visual and auditory blockage. CT/MRI imaging and routine acoustic tests were carried out in all subjects. PET revealed that an increased uptake of 18F-FDG at left med-temporal lobe (primary auditory center, PAC) present exclusively in tinnitus, regardless the side of hearing hallucination. Significant asymmetry was noted between left and right PAC, but not at other cortex area. While control cases showed no asymmetric uptake between two hemispheres. The abnormal PAC uptake did not respond to external pure sound stimulus, nor did it relate to the seventy of hearing loss assessed by acoustic tests. No anatomical or morphological alteration could be proven on CT/MRI. In conclusion, PET/18F-FDG objectively revealed an increased metabolic change at left PAC in tinnitus, which is of diagnostic value; and there is evidence suggesting tinnitus is most likely induced by a functional change in the brain.

个性化护理干预对肿瘤患者^(18)F-FDG PET/CT检查质量及图像质量的影响研究

目的探讨个性化护理干预对肿瘤患者^(18)F-FDG PET/CT检查质量及图像质量的影响。方法选取2020年1月至2022年12月在广西医科大学附属肿瘤医院行^(18)F-FDG PET/CT检查的100例肿瘤患者作为研究对象,随机分为观察组和对照组,每组各50例。对照组实施常规护理干预,观察组实施个性化护理干预,比较两组患者的检查质量及图像质量。结果观察组在显像剂外漏(P=0.017)、注射次数(P=0.042)、肠道摄取增高(P=0.046)、声带摄取增高(P=0.037)、眼肌摄取增高(P=0.027)、全身肌肉摄取增高(P=0.037)发生率均低于对照组,差异有统计学意义(P<0.05)。结论对肿瘤患者在^(18)F-FDG PET/CT显像检查过程中实施个性化护理干预,有利于提高检查质量和图像质量,使患者和医护人员共同获益,值得推广。

全方位管理干预对行^(18)F-FDG PET/CT检查胃癌患者心理状态及满意度影响

目的:探讨全方位管理干预在进行^(18)F脱氧葡萄糖(FDG)正电子发射型计算机断层显像(PET)/CT检查胃癌患者中的应用效果.方法:选取2021年10月至2022年10月本院收治的123例胃癌患者为研究对象.根据干预方案将其分为研究组(n=64,采用全方位管理干预)和对照组(n=59,采用常规干预).采用统计学方法,比较两组患者图像合格率、患者心理状态和检查满意度.结果:研究组患者图像一次合格率高于对照组,差异有统计学意义(P<0.05).研究组患者情绪平稳度、依从性、生活习惯、沟通能力、自我约束评分高于对照组,差异均有统计学意义(P<0.05).研究组患者检查前、中、后检查满意度均高于对照组,差异均有统计学意义(P<0.05).结论:在胃癌患者^(18)F-FDG PET/CT检查中应用全方位管理干预,可改善患者心理状态及满意度,利于诊断检查工作的开展.

^(18)F-FDG PET-CT联合MRI对于前列腺癌的临床诊断价值

目的探讨^(18)F-FDG PET-CT联合MRI在前列腺癌诊断及分期中的应用价值。方法选取我院2020年8月—2023年8月期间收治的70例疑似前列腺癌患者为研究对象,纳入患者均接受PET-CT及MRI检查,以病理穿刺活检结果作为诊断金标准,评价二者联合诊断的诊断效能。结果70例疑似前列腺癌患者,经过穿刺活检确诊为前列腺癌50例,前列腺良性病变20例;其中MRI诊断出前列腺癌43例,前列腺良性病变27例;PET-CT诊断出前列腺癌48例,前列腺良性病变23例;PET-CT+MRI诊断出前列腺癌51例,前列腺良性病变19例;PET-CT联合MRI诊断的准确度、敏感度高于单一诊断,漏诊率低于单一诊断(P<0.05),而二者联合诊断前列腺癌的特异度及误诊率无统计学差异(P>0.05)。病理分期诊断上,50例经病理活检确诊为前列腺癌的患者,低分化腺癌13例;T期40例(T1~T2期11例,T3期22例,T4期7例),N期10例;对于前列腺癌T分期评价,PET-CT+MRI检查的准确度高于单一检查(P<0.05)。结论PET-CT联合MRI相比二者任一检查对前列腺癌及其病理分期具有较高的诊断效能,可减少漏诊率及误诊率。

基于图拉普拉斯正则化的PET图像核重建方法

正电子发射断层成像(Positron Emission Tomography,PET)在很多疾病的早期诊断中有重要的作用,PET图像重建的难点之一是如何在保持重建图像中病灶边缘特性的同时具有良好的去噪性能.针对此问题,本文提出了一种结合图拉普拉斯正则化和深度图像先验的PET图像核重建方法 .设计了改进的U-net神经网络,将PET前向投影模型中的核系数表示为神经网络的输出;通过先验图像构建图拉普拉斯矩阵,重建问题被建模为基于神经网络的带图拉普拉斯正则化项的最大似然函数优化问题.利用优化转移方法导出了收敛的迭代重建算法,每一次迭代包括由核重建方法更新图像和利用神经网络更新核系数两个步骤.仿真和临床实验结果表明,本文提出的方法在不同的指标下都有更好的重建效果,优于已有核重建方法以及最新的基于深度系数先验的重建方法 .