Thoracic Ganglioneuroma: A Rare Neural Tumor (Case Report and Literature Review)

Ganglioneuroma is an extremely rare tumor that is derived from neural crest. Many ganglioneuroma cases are detected incidentally unless they are large enough to cause compressive symptoms. We report an 18-year-old patient with posterior mediastinal ganglioneuroma which was abutting the descending aorta. The patient underwent successful resection by thoracoscopic approach and was followed up for one year with no complications. In summary, a detailed review with experts in both radiology and pathology is mandated to diagnose these tumors. Informed consent was obtained from the patient.

An Adapted Convolutional Neural Network for Brain Tumor Detection

In medical imaging, particularly for analyzing brain tumor MRIs, the expertise of skilled neurosurgeons or radiologists is often essential. However, many developing countries face a significant shortage of these specialists, which impedes the accurate identification and analysis of tumors. This shortage exacerbates the challenge of delivering precise and timely diagnoses and delays the production of comprehensive MRI reports. Such delays can critically affect treatment outcomes, especially for conditions requiring immediate intervention, potentially leading to higher mortality rates. In this study, we introduced an adapted convolutional neural network designed to automate brain tumor diagnosis. Our model features fewer layers, each optimized with carefully selected hyperparameters. As a result, it significantly reduced both execution time and memory usage compared to other models. Specifically, its execution time was 10 times shorter than that of the referenced models, and its memory consumption was 3 times lower than that of ResNet. In terms of accuracy, our model outperformed all other architectures presented in the study, except for ResNet, which showed similar performance with an accuracy of around 90%.

Advances on Tumor Image Segmentation Based on Artificial Neural Network

Image technology is applied more and more to help doctors to improve the accuracy of tumor diagnosis as well as researchers to study tumor characteristics. Image segmentation technology is an important part of image treatment. This paper summarizes the advances of image segmentation by using artificial neural network including mainly the BP network and convolutional neural network (CNN). Many CNN models with different structures have been built and successfully used in segmentation of tumor images such as supervised and unsupervised learning CNN. It is shown that the application of artificial network can improve the efficiency and accuracy of segmentation of tumor image. However, some deficiencies of image segmentation by using artificial neural network still exist. For example, new methods should be found to reduce the cost of building the marked data set. New artificial networks with higher efficiency should be built.

Automatic Segmentation of Liver Tumor in CT Images with Deep Convolutional Neural Networks

Liver tumors segmentation from computed tomography (CT) images is an essential task for diagnosis and treatments of liver cancer. However, it is difficult owing to the variability of appearances, fuzzy boundaries, heterogeneous densities, shapes and sizes of lesions. In this paper, an automatic method based on convolutional neural networks (CNNs) is presented to segment lesions from CT images. The CNNs is one of deep learning models with some convolutional filters which can learn hierarchical features from data. We compared the CNNs model to popular machine learning algorithms: AdaBoost, Random Forests (RF), and support vector machine (SVM). These classifiers were trained by handcrafted features containing mean, variance, and contextual features. Experimental evaluation was performed on 30 portal phase enhanced CT images using leave-one-out cross validation. The average Dice Similarity Coefficient (DSC), precision, and recall achieved of 80.06% ± 1.63%, 82.67% ± 1.43%, and 84.34% ± 1.61%, respectively. The results show that the CNNs method has better performance than other methods and is promising in liver tumor segmentation.

Contrast-enhanced ultrasonography parameters in neural network diagnosis of liver tumors

AIM:To study the role of time-intensity curve(TIC) analysis parameters in a complex system of neural networks designed to classify liver tumors.METHODS:We prospectively included 112 patients with hepatocellular carcinoma(HCC)(n = 41),hypervascular(n = 20) and hypovascular(n = 12) liver metastases,hepatic hemangiomas(n = 16) or focal fatty changes(n = 23) who underwent contrast-enhanced ultrasonography in the Research Center of Gastroenterology and Hepatology,Craiova,Romania.We recorded full length movies of all contrast uptake phases and post-processed them offline by selecting two areas of interest(one for the tumor and one for the healthy surrounding parenchyma) and consecutive TIC analysis.The difference in maximum intensities,the time to reaching them and the aspect of the late/portal phase,as quantified by the neural network and a ratio between median intensities of the central and peripheral areas were analyzed by a feed forward back propagation multi-layer neural network which was trained to classify data into five distinct classes,corresponding to each type of liver lesion.RESULTS:The neural network had 94.45% training accuracy(95% CI:89.31%-97.21%) and 87.12% testing accuracy(95% CI:86.83%-93.17%).The automatic classification process registered 93.2% sensitivity,89.7% specificity,94.42% positive predictive value and 87.57% negative predictive value.The artificial neural networks(ANN) incorrectly classified as hemangyomas three HCC cases and two hypervascular metastases,while in turn misclassifying four liver hemangyomas as HCC(one case) and hypervascular metastases(three cases).Comparatively,human interpretation of TICs showed 94.1% sensitivity,90.7% specificity,95.11% positive predictive value and 88.89% negative predictive value.The accuracy and specificity of the ANN diagnosis system was similar to that of human interpretation of the TICs(P = 0.225 and P = 0.451,respectively).Hepatocellular carcinoma cases showed contrast uptake during the arterial phase followed by wash-out in the portal and first seconds of the late phases.For the hypovascular metastases did not show significant contrast uptake during the arterial phase,which resulted in negative differences between the maximum intensities.We registered wash-out in the late phase for most of the hypervascular metastases.Liver hemangiomas had contrast uptake in the arterial phase without agent wash-out in the portallate phases.The focal fatty changes did not show any differences from surrounding liver parenchyma,resulting in similar TIC patterns and extracted parameters.CONCLUSION:Neural network analysis of contrastenhanced ultrasonography-obtained TICs seems a promising field of development for future techniques,providing fast and reliable diagnostic aid for the clinician.

Detection and Classification of Brain Tumor Based on Multilevel Segmentation with Convolutional Neural Network

Magnetic Resonance Imaging (MRI) is an important diagnostic technique for early detection of brain Tumor and the classification of brain Tumor from MRI image is a challenging research work because of its different shapes, location and image intensities. For successful classification, the segmentation method is required to separate Tumor. Then important features are extracted from the segmented Tumor that is used to classify the Tumor. In this work, an efficient multilevel segmentation method is developed combining optimal thresholding and watershed segmentation technique followed by a morphological operation to separate the Tumor. Convolutional Neural Network (CNN) is then applied for feature extraction and finally, the Kernel Support Vector Machine (KSVM) is utilized for resultant classification that is justified by our experimental evaluation. Experimental results show that the proposed method effectively detect and classify the Tumor as cancerous or non-cancerous with promising accuracy.

Similarity on neural stem cells and brain tumor stem cells in transgenic brain tumor mouse models

Although it is believed that glioma is derived from brain tumor stem cells, the source and molecular signal pathways of these cells are still unclear. In this study, we used stable doxycycline-inducible transgenic mouse brain tumor models （c-myc/SV40Tag＋/Tet-on＋） to explore the malignant trans- formation potential of neural stem cells by observing the differences of neural stem cells and brain tumor stem cells in the tumor models. Results showed that chromosome instability occurred in brain tumor stem cells. The numbers of cytolysosomes and autophagosomes in brain tumor stem cells and induced neural stem cells were lower and the proliferative activity was obviously stronger than that in normal neural stem cells. Normal neural stem cells could differentiate into glial fibrillary acidic protein-positive and microtubule associated protein-2-positive cells, which were also negative for nestin. However, glial fibrillary acidic protein/nestin, microtubule associated protein-2/nestin, and glial fibrillary acidic protein/microtubule associated protein-2 double-positive cells were found in induced neural stem cells and brain tumor stem cells. Results indicate that induced neural stem cells are similar to brain tumor stem cells, and are possibly the source of brain tumor stem cells.

Application of Convolutional Neural Networks in Classification of GBM for Enhanced Prognosis

The lethal brain tumor “Glioblastoma” has the propensity to grow over time. To improve patient outcomes, it is essential to classify GBM accurately and promptly in order to provide a focused and individualized treatment plan. Despite this, deep learning methods, particularly Convolutional Neural Networks (CNNs), have demonstrated a high level of accuracy in a myriad of medical image analysis applications as a result of recent technical breakthroughs. The overall aim of the research is to investigate how CNNs can be used to classify GBMs using data from medical imaging, to improve prognosis precision and effectiveness. This research study will demonstrate a suggested methodology that makes use of the CNN architecture and is trained using a database of MRI pictures with this tumor. The constructed model will be assessed based on its overall performance. Extensive experiments and comparisons with conventional machine learning techniques and existing classification methods will also be made. It will be crucial to emphasize the possibility of early and accurate prediction in a clinical workflow because it can have a big impact on treatment planning and patient outcomes. The paramount objective is to not only address the classification challenge but also to outline a clear pathway towards enhancing prognosis precision and treatment effectiveness.

基于Transformer与注意力机制的肺部肿瘤分割方法

肺部肿瘤的准确分割对于肿瘤的诊断和治疗具有重要作用,然而肺部肿瘤分割中存在病灶与周围组织的对比度低、肿瘤与正常组织易粘连和背景噪声大等问题。针对这些问题,本文提出了一种基于Transformer和注意力机制的肺部肿瘤分割方法。在Transformer编码器阶段引入全局和局部的注意力机制,使得网络可以同时关注全局和局部的上下文信息;在跳跃连接阶段,使用通道优先卷积注意力机制,可以增强复杂病灶的空间感知能力和降低通道维度冗余,从而提高肿瘤的分割精度。在私有数据集GDPH和公共数据集LUNG1上的测试结果表明,本文方法相比其他8种分割方法,Dice指标在两个数据集上表现最优,分别为90.96%和88.18%,可以为临床的诊疗提供可靠辅助。

Upregulation of stromal cell-derived factor-1 alpha/CXCR4 axis-induced migration of human neural progenitors by tumor necrosis factor-alpha and interleukin-8

BACKGROUND： Studies of several animal models of central nervous system diseases have shown that neural progenitor cells （NPCs） can migrate to injured tissues. Stromal cell-derived factor 1 alpha （SDF-la）, and its primary physiological receptor CXCR4, have been shown to contribute to this process. OBJECTIVE： To investigate migration efficacy of human NPCs toward a SDF-1α gradient, and the regulatory roles of tumor necrosis factor-α （TNF-α） and interleukin-8 （IL-8） in SDF-1α/CXCR4 axis-induced migration of NPCs. DESIGN, TIME AND SETTING： An in vitro, randomized, controlled, cellular and molecular biology study was performed at the Laboratory of Department of Cell Biology, Medical College of Soochow University between October 2005 and November 2007. MATERIALS： SDF-1α and mouse anti-human CXCR4 fusion antibody were purchased from R＆D Systems, USA. TNF-αwas purchased from Biomyx Technology, USA and IL-8 was kindly provided by the Biotechnology Research Institute of Soochow University. METHODS： NPCs isolated from forebrain tissue of 9 to 10-week-old human fetuses were cultured in vitro. The cells were incubated with 0, 20, and 40 ng/mL TNF-α, or 0, 20, and 40 ng/mL IL-8, for 48 hours prior to migration assay. For antibody-blocking experiments, cells were further pretreated with 0, 20, and 40 μg/mL mouse anti-human CXCR4 fusion antibody for 2 hours. Subsequently, the transwell assay and CXCR4 blockade experiments were performed to evaluate migration of human NPCs toward a SDF-1α gradient. Serum-free culture medium without SDF-1α served as the negative control. MAIN OUTCOME MEASURES： The transwell assay was performed to evaluate migration of human NPCs toward a SDF-1α gradient, which was blocked by fusion antibody against CXCR4. In addition, CXCR4 expression in human NPCs stimulated by TNF-α and IL-8 was measured by flow cytometry. RESULTS： Results from the transwell assay demonstrated that SDF-1α was a strong chemoattractant for human NPCs （P 〈 0.01）, and 20 ng/mL produced the highest levels of migration. Anti-human CXCR4 fusion antibody significantly blocked the chemotactic effect （P 〈 0.05）. Flow cytometry results showed that treatment with TNF-α and IL-8 resulted in increased CXCR4 expression and greater chemotaxis efficiency of NPCs towards SDF-1α（P 〈 0.01）. CONCLUSION： These results demonstrated that SDF-la significantly attracted NPCs in vitro, and neutralizing anti-CXCR4 antibody could block part of this chemotactic function. TNF-α and IL-8 increased chemotaxis efficiency of NPCs towards the SDF-1αgradient by upregulating CXCR4 expression in NPCs.

中枢神经系统环路在抑郁促进乳腺癌进程中的作用

随着神经科学与肿瘤学的发展,中枢神经系统环路对肿瘤的直接调控作用被逐渐揭示。越来越多的证据表明,靶向情绪相关脑区的治疗,可能在阻断抑郁促进乳腺癌进程中具有很大的潜力,其背后复杂的机制涉及抑郁产生及中枢神经系统环路对肿瘤的调控,然而,该研究领域尚缺乏系统性的总结。本文综述了中枢神经系统环路与抑郁产生、中枢神经系统与外周肿瘤神经联系、交感神经系统调控肿瘤免疫微环境的最新研究进展,系统梳理了中枢神经系统环路在抑郁促进乳腺癌进展中的潜在机理,以期为乳腺癌的综合治疗提出新的解决方法。

Establishment and expression of recombinant human glial cell linederived neurotrophic factor and TNF α receptor in human neural stem cells

Objective:To investigate the interference and expression of human glial cell line-derived neurotrophic factor(hCDNF) and soluble TNF alpha(sTMFRⅠ) receptor genes in neural stem cells and to evaluate the roles of these proteins in the genetic treatment of spinal cord injury.Methods:Full-length of GDNF cDNA(538 bp) and sTMFRⅠcDNA(504 bp) were inserted into the early 1 region of adenovirus genomic DNA respectively and were immediated by the human cytomegalovirus(gene promoter/enhancer). These adenoviruses were propagated in HEK293 cells via homologous recombination for 7-10 days in vivo,then they were used to infect human neural stem ceils.The infection and expression of gene were tested under immunofluorescence.ELISA and Westem-blot after 48 hours.Results:Almost all the cultured cells showed the nestin immunofluorescence positive staining,which was the characteristics of neural stem cell.A great quantity of EGFP and KFP were observed in neural stem cells,which indicated the expression of GDNF and sTMFRⅠ.After transfection of GDNF and sTMFRⅠgenes,many neural stem cells show GFAP and tubulin immunofluorescence positive staining,which meant that most neural stem cells differentiated into neuron at that condition.Conclusions:The infective efficiency of adenovirus is greatly acceptable to neural stem cell,thus adenovirus provide a useful vector for exogenous GDNF and sTMFRⅠgenes expressing in neural stem cells,which is useful for differentiation of neural stem cell.

Neuronal injury and tumor necrosis factor-alpha immunoreactivity in the rat hippocampus in the early period of asphyxia-induced cardiac arrest under normothermia

Low survival rate occurs in patients who initially experience a spontaneous return of circulation after cardiac arrest（CA）. In this study, we induced asphyxial CA in adult male Sprague-Daley rats, maintained their body temperature at 37 ± 0.5°C, and then observed the survival rate during the post-resuscitation phase. We examined neuronal damage in the hippocampus using cresyl violet（CV） and Fluore-Jade B（F-J B） staining, and pro-inflammatory response using ionized calcium-binding adapter molecule 1（Iba-1）, glial fibrillary acidic protein（GFAP）, and tumor necrosis factor-alpha（TNF-α） immunohistochemistry in the hippocampus after asphyxial CA in rats under normothermia. Our results show that the survival rate decreased gradually post-CA（about 63% at 6 hours, 37% at 1 day, and 8% at 2 days post-CA）. Rats were sacrificed at these points in time post-CA, and no neuronal damage was found in the hippocampus until 1 day post-CA. However, some neurons in the stratum pyramidale of the CA region in the hippocampus were dead 2 days post-CA. Iba-1 immunoreactive microglia in the CA1 region did not change until 1 day postCA, and they were activated（enlarged cell bodies with short and thicken processes） in all layers 2 days postCA. Meanwhile, GFAP-immunoreactive astrocytes did not change significantly until 2 days post-CA. TNF-α immunoreactivity decreased significantly in neurons of the stratum pyramidale in the CA1 region 6 hours post-CA, decreased gradually until 1 day post-CA, and increased significantly again 2 days post-CA. These findings suggest that low survival rate of normothermic rats in the early period of asphyxia-induced CA is related to increased TNF-α immunoreactivity, but not to neuronal damage in the hippocampal CA1 region.

Effect of oxymatrine on interferon-gamma and tumor necrosis factor-alpha serum levels in an experimental rat model of autoimmune encephalomyelitis

BACKGROUND： Studies have demonstrated that experimental autoimmune encephalomyelitis （EAE） onset correlates with increased interferon-v （IFN-γ） and tumor necrosis factor-α （TNF-α） expression. Oxymatrine （OM） has been shown to inhibit autoimmune responses, but there are no reports showing that it could prevent the development of EAE. OBJECTIVE： To observe the effect of OM on serum levels of IFN-γ and TNF-α in a rat model of EAE.DESIGN, TIME AND SETTING： A randomized, controlled, animal study was performed at the Experimental Animal Center of Henan Academy of Chinese Medicine and at the Key Disciplines Laboratory Clinical Medicine of Henan Province between July and December 2008. MATERIALS： OM was purchased from Chia-tai Tianqing Pharmaceutical, China; complete Freund＇s adjuvant was purchased from Sigma, USA. METHODS： Forty female Wistar rats were randomly assigned to four groups： EAE model （M）, low-dose OM treatment （OM-L）, high-dose OM treatment （OM-H）, and normal control （N, no immunization）, with 10 rats in each group. EAE was established in the M, OM-L, and OM-H groups following immunization with Guinea pig spinal cord homogenate and complete Freund＇s adjuvant. The M and N groups were intraperitoneally injected with normal saline （6.7 mL/kg per day）, the OM-L group received an intraperitoneal injection of OM （100 mg/kg per day）, and the OM-H group received OM （150 mg/kg per day）. MAIN OUTCOME MEASURES： At 16 days after immunization, the degree of histopathological changes in the spinal cord was assessed by hematoxylin-eosin stanining. Enzyme-linked immunosorbent assay was used to detect serum levels of IFN-γ, and radioimmunoassay was utilized to determine serum TNF-α level. Neurological scores were measured on a daily basis according to a 0-5 scale. RESULTS： Daily injections of OM, both high and low doses, resulted in decreased neurological scores in EAE rats （P〈0.01）, as well as reduced cellular infiltration in the spinal cord and decreased levels of serum IFN-γ and TNF-α （P〈 0.01）. CONCLUSION： OM reduced the onset and severity of EAE, which correlated with decreased IFN-γ and TNF-α expression.

融合CNN与Transformer的MRI脑肿瘤图像分割

为解决卷积神经网络(convolutional neural network,CNN)在学习全局上下文信息和边缘细节方面受到很大限制的问题,提出一种同时学习局语义信息和局部空间细节的级联神经网络用于脑肿瘤医学图像分割。首先将输入体素分别送入CNN和Transformer分支,在编码阶段结束后,采用一种双分支融合模块将2个分支学习到的特征有效地结合起来以实现全局信息与局部信息的融合。双分支融合模块利用哈达玛积对双分支特征之间的细粒度交互进行建模,同时使用多重注意力机制充分提取特征图通道和空间信息并抑制无效的噪声信息。在BraTS竞赛官网评估了本文方法,在BraTS2019验证集上增强型肿瘤区、全肿瘤区和肿瘤核心区的Dice分数分别为77.92%,89.20%和81.20%。相较于其他先进的三维医学图像分割方法,本文方法表现出了更好的分割性能,为临床医生做出准确的脑肿瘤细胞评估和治疗方案提供了可靠依据。

Early rehabilitation improves neurofunctional outcome after surgery in children with spinal tumors

To investigate the effect of early rehabilitation on neurofunctional outcome after surgery in chil- dren with spinal tumors, this study reviewed the medical charts and radiographic records of 70 pediatric patients （1-17 years old） who received spinal tumor surgical removal. The peddiatric patients received rahabilitation treatment at 4 （range, 2-7） days after surgery for 10 （range, 7-23） days. Results from the Modified McCormick Scale, Functional Independence Measure for Chil- dren, American Spinal Injury Association Impairment Scale and Karnofsky Performance Status Scale demonstrated that the sensory function, motor function and activity of daily living of pedi- atric children who received early rehabilitation were significantly improved. Results also showed that tumor setting and level localization as well as patients＇s clinical symptoms have no influences on neurofunctional outcomes.

深度卷积网络模型可自动识别与分割胰腺及其肿瘤:基于3D V-Net

目的探讨基于V-Net的深度卷积神经网络模型在胰腺及其肿瘤自动识别和分割任务中的有效性和可行性。方法回顾性分析2012年5月~2019年11月于上海交通大学医学院附属第一人民医院就诊且经病理证实为胰腺癌的186例患者的增强CT影像资料,经过筛选,共纳入108例胰腺癌病例,随机搜集同期37例正常胰腺病例用于对照,最终共纳入145例数据,构成本研究的数据集。采用五折交叉验证方法,在动脉期CT图像上进行人工标注感兴趣区域(ROI),包括胰腺头颈部、体尾部和肿瘤,通过计算敏感度、特异度、F1分数等指标评估模型对胰腺肿瘤的识别能力,并进行Kappa一致性验证。采用Dice系数定量评估模型的分割能力,并获取可视化结果进一步评估。结果基于V-Net的模型识别胰腺肿瘤的敏感度为0.852、特异度为1.000、阳性预测值为1.000、阴性预测值为0.698,F1分数高达0.920。一致性验证显示,Kappa系数为0.746(P<0.05)。在分割任务中,胰腺肿瘤、胰腺体尾部和胰腺头颈部的Dice系数分别为0.722±0.290、0.602±0.175、0.567±0.200。结论本研究构建基于VNet的深度卷积网络模型,有效完成了胰腺及其肿瘤自动识别与分割,验证了该方法的有效性和可行性,为探索胰腺肿瘤领域人工智能应用提供有力支持。

卷积神经网络在肝癌病理图像诊断中的应用综述

肝癌是一种恶性肝肿瘤,起源于肝细胞.肝癌诊断一直是医学难点问题,也是各领域研究的热点问题,早期确诊肝癌可以降低肝癌的死亡率.组织病理学图像检查是肿瘤学诊断的黄金标准,图像会显示组织切片的细胞和组织结构,可以用于确定细胞类型、组织结构、异常细胞的数量和形态,并评估肿瘤具体情况.本文重点研究了卷积神经网络针对病理图像的肝癌诊断算法,包括肝肿瘤检测、图像分割以及术前预测这3个方面的应用,详细阐述了卷积神经网络各算法的设计思路和相关改进目的及方法,以便为研究人员提供更清晰的参考思路.总结性分析了卷积神经网络算法在诊断中的优缺点,并对未来可能的研究热点和相关难点进行了探讨.

基于卷积神经网络和迁移学习的肿瘤舌象识别研究

目的 探究卷积神经网络Res Net152模型识别肿瘤舌象图片的性能。方法 选取2019年1月至2021年12月于北京中医药大学东直门医院、北京中医药大学东方医院和北京中医药大学第三附属医院采集的5 943张舌图,其中包括肿瘤舌图1 433张,非肿瘤舌图4 510张。经过图像预处理后,随机选取1 000张舌图作为测试集,其余的4 943张作为训练集。通过图像扩增技术将4 943张训练集舌图扩增为54 535张,并输入在ImageNet-2012数据集上经过预训练的卷积神经网络ResNet152模型,以建立舌象自动识别系统。然后将1 000张测试集舌图输入模型,记录识别结果。最后,运用GRAD-CAM技术对测试集中模型正确识别为肿瘤的舌图进行可视化分析,统计和分析模型识别肿瘤舌图重点关注的舌象特征。结果 在测试集中,卷积神经网络ResNet152模型识别肿瘤舌象的分类正确率为85.7%,召回率为84.9%,准确率为85.5%,F_(1)值为85.2%,曲线下面积为91.3%。对正确识别为肿瘤的舌图进行可视化分析,结果显示对模型正确识别肿瘤贡献度最高的舌象特征是瘀斑和裂纹。结论 卷积神经网络ResNet152模型提供了一种非侵入性且高效的肿瘤检测方法,能够辅助肿瘤的诊断,瘀斑和裂纹可能是模型预测肿瘤最关注的两个舌象特征。

基于深度学习的三维肿瘤及器官分割

针对三维医学图像中由于肿瘤或器官的形状、尺度差异较大导致分割精度较低的问题,提出一种端到端的三维全卷积分割模型。首先,设计空洞立方集成模块在不同分辨率阶段实现多尺度集成,增强复杂边界上的识别能力;其次,引入跨阶段上下文融合模块融合浅层和深层特征,促进收敛并更准确地定位目标对象;最后,解码器对来自编码器的特征进行拼接以实现分割。在脑肿瘤分割数据集上,平均Dice相似性系数值达到85.37%;在腹部器官分割数据集上,平均Dice相似性系数值达到83.99%。实验结果表明所提模型在三维肿瘤和器官的分割上具有较高精度。