Accurate classification and segmentation of intracranial aneurysms from 3D point cloud data are critical for computer-aided diagnosis and surgical planning. However, existing point-based deep learning methods suffer from limited feature representation and poor segmentation performance on medical data due to insufficient training samples and complex geometric variations. M-PointNet introduces a novel multi-layer embedded deep learning architecture that significantly enhances the classification and segmentation of intracranial aneurysms through three key innovations: (1) an enhanced PointNet++ with an expanded hierarchical structure for better geometric feature extraction; (2) a multi-layer embedding mechanism that integrates preprocessed and resampled point cloud data at multiple hierarchical levels to enrich feature representation; and (3) a deep supervision strategy with auxiliary output layers to accelerate convergence and improve performance. Experiments on the IntrA dataset demonstrate that M-PointNet achieves 91.96% accuracy and a 0.923 F1-score in classification, surpassing baseline by 5.27% and 3.0%, respectively. For segmentation, it attains 83.85% IoU and 90.25% DSC for aneurysm regions and 95.81% IoU and 97.82% DSC for vessel regions. Additionally, its generalization capability is validated by a 92.8% accuracy on the ModelNet40 dataset. M-PointNet effectively addresses the challenges of medical point cloud analysis, achieving state-of-the-art performance in intracranial aneurysms classification and segmentation while maintaining robust cross-domain generalization.

Finger vein recognition faces two major challenges: the degradation of recognition performance due to varying lighting conditions during image acquisition and the limitation of available training data. This paper proposes FV-Res50ED, a novel model designed to address both issues. First, we develop a Focus-ROI method to isolate key finger vein regions, excluding irrelevant background areas. Second, we propose a Dynamic Contrast-Limited Adaptive (DCLA) enhancement method. It dynamically adjusts the contrast limit using local image features, efficiently reducing blur and illumination-induced artifacts. Third, we introduce a Triplet Angular Adaptive Margin (TAAM) loss, which integrates angular loss into triplet loss, utilizing cosine similarity to strengthen feature discrimination. Furthermore, TAAM adopts a adaptive learnable margin to enhance the discriminative power of features. Experimental results on three public datasets show that FV-Res50ED achieves recognition accuracies of 99.69

Thyroid carcinoma (THCA) exhibits molecular heterogeneity, necessitating robust biomarkers for precise diagnosis and mechanistic insights. In this study, we integrated the transcriptomic data of THCA and normal thyroid tissues from The Cancer Genome Atlas and Genotype-Tissue Expression (GTEx) databases using ComBat-based batch correction, identifying 5,573 differentially expressed genes. Functional enrichment showed that these genes were associated with dysregulation in immune-microenvironment interactions and post-transcriptional regulatory pathways. After a comparison of seven machine-learning algorithms, LightGBM was selected for its high discrimination performance and interpretability. Using this algorithm, we obtained an 8-gene diagnostic panel comprising a glycosyltransferase (GALNT7), pseudogenes (HIRAP1, SKP1P1), miRNAs (MIR331, MIR93), and novel transcripts. siRNA-mediated knockdown of GALNT7 and SKP1P1 significantly attenuated the proliferative and migratory abilities and clonogenicity of the THCA cell line TPC-1 in vitro. Immune correlation analysis revealed the tumor-specific suppression of stromal components by GALNT7 and normal tissue-specific lymphoid regulation by SKP1P1, suggesting the different roles of these components in microenvironment remodeling. Knockdown of HIRAP1, MIR331, and MIR93 had no statistically significant effects on these oncogenic phenotypes. However, their high predictive value in the computational model suggests they may serve as effective biomarkers for tumor classification, independent of their functional roles in tumorigenesis. These findings highlight the importance of orthogonal experimental validation in linking computational biomarker discovery with biological causality. This study suggests that GALNT7 and SKP1P1 are promising diagnostic biomarkers and therapeutic targets for THCA, with dual roles in tumor-specific signaling and immune microenvironment modulation.

Intracranial aneurysms are relatively common and life-threatening conditions, making precise segmentation during early diagnosis crucial. However, the challenges of poor imaging quality and high noise levels often result in unclear aneurysm edges. Additionally, the varying sizes of aneurysms further complicate accurate segmentation. To address these issues, we propose a Multiscale and Edge-guided enhanced 3D deep learning model. First, the asymmetrically larger network with enhanced hierarchical feature representation effectively captures subtle image features, thereby improving the localization of anatomical structures. Second, the multi-scale feature fusion mechanism within the encoder improves feature diversity and edge information, enhancing segmentation precision for aneurysms of different sizes. Finally, the edge-guided attention technique within the decoder combines local features with predicted heatmaps to extract comprehensive edge information. The experimental results demonstrate that the model outperforms general models in five key metrics on the internal dataset. External dataset testing confirms its adaptability and robustness across data from different acquisition protocols and hardware configurations. Clinical trials have further validated its practicality, assisting radiologists in more accurate intracranial aneurysm diagnosis.

This paper introduces an innovative framework for detecting pulmonary nodules in low-dose CT scans, utilizing a 3D single-stage detection model combined with an adversarial data augmentation strategy. Although conventional U-Net and SSD-inspired 3D detection models are effective at learning detailed nodule features, the severe class imbalance between nodules and the surrounding lung tissue often limits their performance in achieving both high sensitivity and specificity. To overcome this challenge, a 3D conditional generative adversarial network (3D-cGAN) augmentation module is introduced, aiming to synthesize realistic nodules precisely at specified locations within back-ground regions. This novel augmentation approach significantly mitigates the class imbalance problem, thereby improving the model's effectiveness in differentiating nodules from background tissues. Compared to traditional two-stage detection systems, our integrated single-stage model demonstrates superior efficiency and improved detection accuracy on the publicly available LUNA16 dataset derived from LIDC-IDRI.

Alzheimer’s disease is a neurodegenerative disease that seriously threatens the life and health of the elderly. This study used three-dimensional lightweight neural networks to classify the stages of Alzheimer’s disease and explore the relationship between the stages and the variations of brain tissue. The study used CAT12 to preprocess magnetic resonance images of the brain and got three kinds of preprocessed images: standardized images, segmented standardized gray matter images, and segmented standardized white matter images. The three kinds of images were used to train four kinds of three-dimensional lightweight neural networks respectively, and the evaluation metrics of the neural networks are calculated. The accuracies of the neural networks for classifying the stages of Alzheimer’s disease (cognitively normal, mild cognitive impairment, Alzheimer’s disease) in the study are above 96%, and the precisions and recalls of classifying the three stages are above 94%. The study found that for the classification of cognitively normal, the best classification results can be obtained by training with the segmented standardized gray matter images, and for mild cognitive impairment and Alzheimer’s disease, the best classification results can be obtained by training with the standardized images. The study analyzed that in the process of cognitively normal to mild cognitive impairment, variations in the segmented standardized gray matter images are more obvious at the beginning, while variations in the segmented standardized white matter images are not obvious. As the disease progresses, variations in the segmented standardized white matter images tend to become more significant, and variations in the segmented standardized gray matter images and white matter images are both significant in the development of Alzheimer’s disease.

This study proposes EMBNet, a lightweight finger vein recognition method based on MobileNetV2 architecture, to address the insufficient recognition accuracy of conventional finger vein algorithms in complex scenarios. Specifically, we significantly improve feature extraction capability by incorporating a Convolutional Block Attention Module (CBAM) and Swish activation function into MobileNetV2. Extensive comparative experiments on three benchmark datasets demonstrate that EMBNet achieves recognition rates of 98.90%, 99.04%, and 98.98 % respectively, representing a 2.3 percentage point improvement over the original MobileNetV2. The model achieves an 11.3% reduction in inference time compared to state-of-the-art methods. EMBNet maintains lightweight characteristics while delivering high recognition accuracy, providing a practical solution for finger vein recognition on mobile and embedded devices.

Intracranial aneurysm (IA) is a serious threat to human health and can lead to subarachnoid hemorrhage and other serious consequences. If IAs can be detected in advance and treated before rupture, it will greatly reduce the harm of IAs to patients. Since brain arteries are 3D (three-dimensional) structures, point cloud methods can directly process 3D data, which is crucial for tasks that require spatial understanding, such as object detection. The 3D point cloud object detection methods Point-Voxel Feature Set Abstraction (PV-RCNN), Part-Aware and Part-Aggregation (PartA2), and Sparsely Embedded Convolutional Detection (SECOND) were applied to the detection of IAs. The object detection model was trained and tested on the public dataset IntrA. The results indicate that the trained model can be used for the detection and location of IAs and high recall values have been obtained on the testing set. This work also provides important metrics for evaluating object detection models, including average precisions (APs), recall values, and object detection results on the testing set, that is, predicted bounding boxes and corresponding confidence scores. In terms of detection results, the IA detection results of PV-RCNN are the best of these three methods by leveraging both point cloud and image information. The object detection method of 3D point cloud can be integrated into the medical imaging post-processing system and can be used as a subsequent module of the 3D reconstruction module.

Context:Kidney renal clear-cell carcinoma (KIRC) is a malignant tumor. At an early stage, KIRC patients may experience only mild fever and fatigue or even no symptoms, and these early nonspecific indications can delay treatment. Neurotransmitters and their receptors may be very useful in determining tumorigenesis and predicting metastasis.Objective:The study intended to investigate the predictive value of neurotransmitter receptor-related genes (NRRGs) using public KIRC data, by determining the biological processes that implicate the prognostic NRRGs and establishing a predictive NR-related risk model, to provide an empirical basis for identifying and treating KIRC patients.Design:The research team performed a genetic case-control study.Setting:The study took place at Research Center of Health, Big Data Mining and Applications, Wannan Medical College, Wuhu, China.Methods:The research team: (1) obtained the transcriptome data related to KIRC from the Cancer Genome Atlas (TCGA) and ArrayExpress databases; (2) developed the differentially expressed NRRGs (DENRRGs) by identifying the NRRGs that intersected with DEGs in KIRC and normal samples; (3) carried out functional enrichment analyses of the DENRRGs; (4) screened the characteristic genes of the DENRRGs using machine learning; (5) created a predictive model using multivariate Cox analyses of the distinctive genes; (6) obtained independent prognostic factors for KIRC patients and established a nomograph model; (7) investigated the sensitivity of KIRC patients to therapeutic agents to examine the variations in immunological features between high-risk and low-risk individuals.Results:Differential analysis found that 115 NRRGs intersected with 5275 DEGs to provide 52 DENRRGs. Functional enrichment showed that DENRRGs were mainly involved in signal transduction in the nervous system. The machine learning on the 52 DENRRGs filtered out nine characteristic genes. Subsequently, the research team found eight prognostic biomarkers-histamine receptor H2 (HRH2), gamma-aminobutyric acid (GABA) receptor subunit epsilon (GABRE), cholinergic receptor nicotinic delta subunit (CHRND), glutamate receptor ionotropic subunit 2D (GRIN2D), glutamate metabotropic receptor 4 (GRM4), glycine receptor alpha 3 (GLRA3), cholinergic receptor nicotinic beta 4 subunit (CHRNB4), and cholinergic receptor muscarinic-1 (CHRM1)-and established a predictive model. Furthermore, the team precisely predicted the KIRC patients' prognoses using a nomogram that combined their ages, risk scores, and M stages. The infiltration levels of 21 immune cells also significantly differed between the high-risk and low-risk groups, with neutrophils having a significant positive correlation with GABRE and HRH2 and a significant negative correlation with CHRNB4 and GRM4. Finally, the 50% inhibitory concentration (IC50) values for various drugs, such as 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR), 8-hydroxy-7-(6-sulfonaphthalen-2-yl)diazenyl-quinoline-5-sulfonic acid (NSC-87877), Sunitinib, c-Jun N-terminal kinase (JNK) inhibitor VIII, and tanespimyci (X17.AAG) were significantly lower for high-risk group.Conclusions:By studying the relevance of biomarkers to the immunological microenvironment of KIRC, the current research team was able to propose a new predictive model for KIRC based on NRRGs, to offer a novel viewpoint for investigating KIRC. The study's results suggest new avenues for research into the pathophysiology and therapy of KIRC. Determining the precise molecular processes by which predictive biomarkers regulate KIRC requires further evidence and analysis.

Mr.Vc is a database of curated Vibrio cholerae transcriptome data and annotated information. The main objective is to facilitate the accessibility and reusability of the rapidly growing Vibrio cholerae omics data and relevant annotation. To achieve these goals, we performed manual curation on the transcriptome data and organized the datasets in an experiment-centric manner. We collected unknown operons annotated through text-mining analysis that would provide more clues about how Vibrio cholerae modulates gene regulation. Meanwhile, to understand the relationship between genes or experiments, we performed gene co-expression analysis and experiment-experiment correlation analysis. In additional, functional module named “Interactions” which dedicates to collecting experimentally validated interactions about Vibrio cholerae from public databases, MEDLINE documents and literature in life science journals. To date, Mr.Vc v2, which is significantly increased from the previous version, contains 107 microarray experiments, 106 RNA-seq experiments, and 3 Tn-seq projects, covering 56,839 entries of DEGs (Differentially Expressed Genes) from transcriptomes and 7,463 related genes from Tn-seq, respectively. and a total of 270,129 gene co-expression entries and 11,990 entries of experiment-experiment correlation was obtained, in total 1,316 entries of interactions were collected, including 496 protein-chemical signaling molecule interactions, 472 protein–protein interactions, 306 TF (Transcription Factor)-gene interactions and 42 Vibrio cholerae-virus interactions, most of which obtained from 402 literature through text-mining analysis. To make the information easier to access, Mr.Vc v2 is equipped with a search widget, enabling users to query what they are interested in. Mr.Vc v2 is freely available at http://mrvcv2.biownmc.info.

目的 利用肿瘤基因组图谱(TCGA)数据库筛选差异表达mRNA、miRNA和lncRNA分子标签,构建预后相关的核心ceRNA调控子网络,探索其在乳腺癌预后相关生物标志物筛选中的作用.方法 从TCGA下载乳腺癌患者的测序数据和临床数据,利用R包"Deseq2"和单因素COX比例风险回归模型,分别筛选差异表达且与预后相关的mR-NA、miRNA和lncRNA分子标签,基于预测性数据库构建ceRNA网络,并对网络中的核心mRNA进行功能注释和免疫相关性分析.结果 筛选出3个mRNA(KLF11、EDA、STXBP1)、1个lncRNA(XIST)和2个miRNA(hsa-miR-130a-3 p、hsa-miR-195-5 p)组成的核心ceRNA子网络,生存分析显示miRNA低表达、mRNA和lncRNA高表达,总生存期显著降低.结论 核心ceRNA子网络中的6个RNA相互调节,可能通过免疫系统在乳腺癌发生发展中起重要作用,对预后产生影响.

In this study, deep learning and triplet loss function methods are used for finger vein verification research, and the model is trained and validated between different kinds of datasets including FV-USM, HKPU, and SDUMLA-HMT datasets. This work gives the accuracy and other evaluation indexes of finger vein verification calculated for different training-validation set combinations and gives the corresponding ROC curves and AUC values. The accuracy of the best result has reached 98%, and all the ROC AUC values are above 0.98, indicating that the obtained model can identify the finger veins well. Since the experiments are cross-validated between different kinds of datasets, the model has good adaptability and applicability. From the experimental results, it is also found that the model trained on the dataset that is more difficult to be distinguished will be a better and more robust model.

With rapid development of economy and technology, people’s life style has changed significantly. Specifically, physical activity has been greatly reduced while the consumption of fast food and processed food has increased, and then the obesity has become more and more common and serious problem. In order to solve this problem, a heat monitoring and management system based on mobile terminals is proposed and realized. By embedding the heat monitoring and management module into mobile terminals, the proposed system allows users to facilitate their weight management regardless of time and space.

在联邦推荐系统中,各客户端能否获得满意的推荐效果不仅取决于自身参与模型训练的数据,也取决于进行联合建模时其他客户端提供的数据.然而出于对自身数据安全的保护且数据获取不易,各客户端倾向于尽可能少地提供自身数据,期望其他客户端提供更多的数据来获得较好的推荐效果.文章首先使用不完全信息博弈模型对客户端之间的交互行为进行分析,接着引入满足均衡概念对该模型进行解释,假定各客户端均有一个预期推荐效果,当所有客户端都达到预期推荐效果时,即该博弈达到均衡.该文提出一种基于均衡学习的迭代算法,客户端通过分析当前推荐效果动态调整本地模型训练的数据,最终使各客户端均达到满足状态.理论分析和实验仿真表明,所提算法可以使各客户端均达到满足均衡,完成收敛.

The rapid development of big data technology has attracted a variety of sectors, including tertiary education. The purpose of this paper is to construct a precision teaching mode based on big data technology in order to improve teaching quality and further promote education and teaching reform. The proposed mode, based on the theory of precision teaching in colleges and universities as well as the intrinsic properties of big data teaching activities, describes five procedures for analyzing learning situations, determining teaching goals, preparing teachers, and evaluating teachers. When the big data-based precision teaching mode is applied to the “Python Language Programming” course, the results show that students are more satisfied with the design of the teaching and more efficient in learning. It is believed that this mode will significantly improve students' academic performance and their ability to work independently and collaboratively as a result of more frequently online and offline interactions between teachers and students.

糖尿病性视网膜病变是一种难以诊断、高风险的致盲性疾病.针对人工对图像特征提取困难、分类准确性差、耗费时间长的问题,采用卷积神经网络构建糖尿病性视网膜病变自动分类器具有重要的临床价值.方法:本文针对已收集好的彩色眼底图像,通过对图像的清洗、扩增、归一化构建糖尿病性视网膜病变数据集.利用VGG16与FCN的优点将其结合,将全连接层改造为卷积层,构建新的糖尿病性视网膜病分类模型.将ImageNet充分训练好的VGG16网络模型参数作为本文模型初始化参数,送入已改造的神经网络模型提取特征,最后输出分类结果.结果:实验结果表明,本文提出的深度学习分类方法的准确率与损失值均优于传统同类别的卷积神经网络分类算法,对临床诊断参考有重要的意义.结论:本文利用的方法对解决数据分布不均衡和过拟合的问题有一定的促进作用,具有较好的鲁棒性.

在计算机辅助眼底图像视网膜血管分割中,基于匹配滤波算法的应用非常广泛.而传统匹配滤波器算法存在分割细小血管效果较差、噪声多以及视盘干扰等问题.本文提出一种相似度滤波算法的眼底图像视网膜血管分割方法.首先用多层阈值和水平集算法提取视盘干扰区域,利用高斯模糊去除视盘干扰区域.然后采用相似度滤波运算对去除视盘干扰的彩色眼底图像进行处理.最后,将余弦相似度图进行二值化后与余弦相似度加强图进行区域连通性判断,实现眼底图像视网膜血管分割.结果 表明,该算法能较好地分割细小血管以及去除视盘干扰,能更为准确地提取眼底图像视网膜血管.

随着信息技术的快速发展,物联网技术已经成为医疗模式革新的重要推动力,健康医疗行业对掌握医疗物联网的人才有很大的需求量.《医疗物联网开发与实训》课程以健康医疗领域现实需求为基础,以学生能力培养为主线,围绕综合型项目开发进行课程设计与实施.实践表明,以多种医疗应用场景下物联网工程项目开发为驱动的课程教学模式,有利于提高学生学习的积极性和主动性,使学生真正做到学以致用.

目的 探讨面向线上线下混合实验教学模式在医学院校非医学专业的专业导论课程实验教学中的应用效果,以期改进导论课程实验教学模式的缺陷,提升专业导论的实验教学实效.方法 在我校医学信息学院2018级信息管理与信息系统和医学信息工程两个专业学生的《医学信息导论》课程中采用面向线上线下的混合实验教学模式,针对不同的学习形式、内容、方法进行有机的整合,将线下实验教学与线上自学、教学的方法相结合,且考核方法采用多元化、多角度的形成性评价方式.最后,通过考试结果分析、调查问卷和实际访谈的方式了解混合实验教学模式和形成性评价的教学效果.结果 通过成绩分析表得出采用混合实验教学模式的班级考试结果优秀率和良好率均高于采用传统教学模式的班级;同时通过问卷调查分析,98.93％的学生认为基于"SPOC+任务驱动"的混合实验教学模式可以加强师生间的互动,活跃课堂气氛,并有助于培养学生的学习兴趣;96.77％的学生认为混合实验教学模式可以提高学生自主学习的能力;97.85％的学生认为有必要增加课堂学习的过程性评价.结论 面向线上线下的混合实验教学模式,为学生提供了一个"自主、共享、协作"的学习平台,有利于提高学生学习的兴趣和增强学生学习的思维能力,为其他同类课程的实验教学方式提供了模式提供了一种参考思路和方法.

通过量表收集芜湖地区6所大学的大学生积极心理品质状况及影响因素数据,构建样本数据库;使用Xgboost(eXtreme Gradient Boosting)算法模型分析样本数据,并与决策树、SVM(support vector machines)和随机森林算法进行对比.实验表明,Xgboost算法的预测精度最高,其决定系数最大值为0.93,说明了预测的有效性.同时发现家庭环境、学习成绩、每周锻炼时间、社团活动和年级等影响因素对模型贡献较高.本研究对预测大学生积极心理品质状况及影响因素分析有一定的参考价值.