Accurate polyp segmentation in colonoscopy images is essential for early colorectal cancer detection but remains a challenging problem due to the limitations in existing methods for optimizing boundary features and aligning cross-level representations. Specifically, the indistinct polyp boundaries and scale variations across different feature levels pose significant challenges for segmentation accuracy. To address these issues, we propose a dual domain collaborative network (DDCNet) that introduces two novel modules: a frequency context enhancement module (FCEM), which operates in the frequency domain to refine high- and low-frequency features, and a cross-level shift-recalibrated fusion module (CSFM), which improves multi-scale feature alignment in the spatial domain. The FCEM improves boundary precision by adaptively refining high-frequency boundary features and enhancing low-frequency contextual information, while the CSFM mitigates cross-level feature misalignment by dynamically recalibrating multi-scale features throughout the encoder-decoder architecture. Additionally, we design a hybrid loss function that integrates boundary, cross-entropy, and frequency consistency losses to further boost segmentation performance. Experimental results on three benchmark datasets (Kvasir-SEG, CVC-ClinicDB, and CVC-ColonDB) demonstrate that DDCNet achieves state-of-the-art performance, with Dice coefficients of 0.9343, 0.9447, and 0.8155, respectively. These results represent improvements of 1.0%-1.5% over the best existing methods. Ablation studies further validate the individual contributions of FCEM, CSFM, and the hybrid loss function. Additionally, we compared the proposed loss function with three commonly used functions.

Scoliosis significantly impacts quality of life, highlighting the need for effective early scoliosis screening (SS) and intervention. However, current SS methods often involve physical contact, undressing, or radiation exposure. This study introduces an innovative, non-invasive SS approach utilizing a monocular RGB camera that eliminates the need for undressing, sensor attachment, and radiation exposure. We introduce a novel approach that employs Parameterized Human 3D Reconstruction (PH3DR) to reconstruct 3D human models, thereby effectively eliminating clothing obstructions, seamlessly integrated with an ISANet segmentation network, which has been enhanced by Multi-Scale Fusion Attention (MSFA) module we proposed for facilitating the segmentation of distinct human trunk and limb features (HTLF), capturing body surface asymmetries related to scoliosis. Additionally, we propose a Swin Transformer-enhanced CMU-Pose to extract human skeleton features (HSF), identifying skeletal asymmetries crucial for SS. Finally, we develop a fusion model that integrates the HTLF and HSF, combining surface morphology and skeletal features to improve the precision of SS. The experiments demonstrated that PH3DR and MSFA significantly improved the segmentation and extraction of HTLF, whereas ST-based CMU-Pose substantially enhanced the extraction of HSF. Our final model achieved a comparable F1 (0.895$\pm$0.014) to the best-performing baseline model, with only 0.79% of the parameters and 1.64% of the FLOPs, achieving 36 FPS-significantly higher than the best-performing baseline model (10 FPS). Moreover, our model outperformed two spine surgeons, one less experienced and the other moderately experienced. With its patient-friendly, privacy-preserving, and easily deployable solution, this approach is particularly well-suited for early SS and routine monitoring.

Objective: To objectively measure Traditional Chinese Medicine (TCM) constitution by moving beyond local features and Body Mass Index (BMI), which only reflects general obesity, through the analysis of overall body shape features. Methods: A deep learning-based parametric 3D human reconstruction approach (PARE) was employed to extract ten body shape parameters (ß0-ß9) from clinical data of 129 subjects. The associations between these parameters and TCM constitution were analyzed using Shapley decomposition to assess their individual contributions. Results: Significant correlations were identified between six TCM constitutions (Phlegm-dampness, Special, Damp-heat, Qi-deficiency, Yin-deficiency, and Yang-deficiency) and body shape parameters. Shapley decomposition quantified the contribution levels of each morphological parameter to the formation of these constitutions. Conclusions: This study provides preliminary insights into the connections between body shape features and TCM constitution using 3D human reconstruction technology, introducing a novel method and perspective for understanding their interplay and advancing the objectification and automation of TCM constitution identification.

Colorectal cancer (CRC) screening is primarily conducted through endoscopy, which remains the gold standard due to its high diagnostic accuracy. However, its invasiveness, operational complexity, and poor patient compliance limit its widespread application. Therefore, intelligent diagnostic approaches based on traditional Chinese medicine (TCM) have gained increasing attention as potential noninvasive alternatives. In this study, an intelligent TCM-assisted diagnostic framework is developed using tongue image analysis for auxiliary CRC diagnosis. Considering the characteristic of limited tongue image samples, the problem is formulated as a few-shot learning task. To address this challenge, a meta-learning-based framework, Meta-Learning with Task-Adaptive Loss and Flexible Affine (MeTALFA), is employed to enhance model adaptability and generalization across tasks. The model integrates task-adaptive loss and dynamic loss-weighting strategies for optimized learning under small-sample conditions. A dataset containing 5,967 tongue images was collected and categorized into healthy and CRC groups. Four meta-learning algorithms, including MAML, MAML++, MeTAL, and MeTALFA, were evaluated under identical experimental settings. Experimental results show that MeTALFA achieves the best diagnostic performance, with an accuracy of$99.67 \% \pm 0.15 {\%}$, precision of 99.57%, recall of 99.77%, and AUC of 98.04%. These findings demonstrate that MeTALFA provides an effective and data-efficient approach for noninvasive CRC screening based on tongue images.

Existing intelligent diagnostic methods in Traditional Chinese Medicine (TCM) heavily rely on clinical experience and rule-based models, which often lack generalizability and overlook medical text data. To address this, we propose a multi-instance multi-label graph neural network model for TCM diagnosis. We frame TCM diagnosis as a multi-instance multilabel problem: each medical record is treated as a bag, where instances are symptom-cluster subgraphs generated via random walks, labeled with syndrome elements (zhengsu). Features are extracted from these sub-instances, aggregated into a bag-level representation, and subsequently mapped to zhengdu labels through a classifier by a graph neural network. Our model achieves a Hamming Loss of 2.58 %, Ranking Loss of 2.56 %, One-Error of 21.55 %, Coverage of 4.565, and Average Precision of 81.55 %, outperforming classic MIML methods including DeepMIML, AttentionMIML, Fast-MIML, and Lnn-MIML across these metrics. These results demonstrate that combining multiinstance multi-label learning with graph neural networks effectively captures symptom-zhengsu relationships, offering a promising direction for intelligent and objective TCM diagnostic research.

Discriminative models, such as Convolutional Neural Networks and Vision Transformers, have driven significant progress in medical image segmentation, their limited generalization ability across diverse datasets remains a key challenge. Consequently, generative models have been increasingly explored for such image-to-image tasks due to their capacity to produce flexible and diverse outputs. Among these, methods based on diffusion models have shown considerable promise. A critical drawback, however, is that existing approaches often fail to adapt these models natively for segmentation, leading to compromised performance. To address this limitation, we introduce MSLD, a novel framework for medical semantic segmentation based on latent diffusion models. Our approach recasts segmentation as a conditional generation problem, where the model is fine-tuned to generate a segmentation mask conditioned on the input image. Extensive experiments on multiple datasets demonstrate that our proposed method achieves high-quality medical image segmentation.

Health and smart healthcare represent a major future trend, in which intelligent traditional Chinese medicine (TCM) plays a significant role. This article reviews recent advances in intelligent assisted diagnosis methods and TCM inspection-based classification tasks. In particular, intelligent TCM inspection diagnosis and its related tasks are the most studied, especially for facial, tongue, eye, and palm diagnoses using the corresponding medical images. Besides, it is mainly divided into body constitution recognition, disease-assisted diagnosis, medical image segmentation tasks by the facial diagnosis and tongue diagnosis. Intelligent TCM assisted diagnostic methods and disease recognition tasks are also included and categorized for overview. From the summary of the reviewed literatures, the main tasks using the medical images are disease classification and recognition, image segmentation, and target detection of disease locations. Finally, the research trends of intelligent TCM four diagnoses and their combination diagnosis are discussed, which can be regarded as the tasks of multi-classification, multi-label, multi-modal fusion, and multi-tasking from the perspective of problem modeling.

In traditional Chinese medicine (TCM), accurately identifying interactions between syndromes and syndrome elements is crucial for precise diagnosis and treatment. To address this challenge, we introduce the Multi-Label Syndrome Classification model based on Graph Attention Networks (ML-SCGAN). The model uses BERT for feature extraction and GAT to capture complex interactions between symptoms and syndrome elements. Experimental results show that ML-SCGAN significantly improves classification, achieving a Micro F1 of 0.631 and a Hamming Loss of 0.033. Compared to baseline models including Binary Relevance, Classifier Chain, One-vs-rest, Label Powerset, HSVM, ML-KNN, CNN, Bi-LSTM, ML-SCGAN shows a marked improvement, outperforming the optimal baseline model HSVM by 12.3% on Micro F1. This approach enhances the efficiency of syndrome classification, promoting the intelligent application of TCM diagnosis.

Introduce the concepts of information extraction and Chinese herbal medicine prescriptions patent. The algorithm design, algorithm framework, and experimental verification of the information extraction were elaborated, indicating that this method can convert semi-structured patent information of Chinese herbal medicine prescriptions into a structured form. Finally, we provide an example of using data mining to analyze the medication patterns in Chinese herbal medicine prescriptions for treating lung cancer.

Medical image recognition methods based on deep learning typically have extremely high requirements for the distribution consistency of source domain data and target domain data. In real-world deployment scenarios, there are often significant differences in target domain data generated by devices from different medical centers and manufacturers. When a medical image recognition model recognizes target domain data that is different from its source domain data distribution, its generalization performance will decrease, greatly reducing the credibility of the medical image recognition model in practical applications.In response to the above issues, this article proposes a parameter adaptive tuning algorithm for medical image recognition model testing. The algorithm uses parallel networks to learn the characteristics and distribution of target domain data, while keeping the parameters of the main network unchanged during adaptation. By optimizing the parameters of the parallel network through backpropagation and gradient descent, it can avoid catastrophic forgetting of the model while also learning the characteristics and distribution of target domain data to a certain extent. In the prediction stage, the fusion network is used to predict the target and improve the generalization performance of the model. This article uses YOLOv5 as the basic model and proposes a parameter adaptive tuning algorithm for medical image recognition model testing. The accuracy of the algorithm on datasets FPPD and DRCOLO has been improved to 94.2% and 97.8%, and the mAP on multiple Disease-Colo video datasets has been improved to 30.5% and 41.7%, verifying the effectiveness of the algorithm.

Background: Epidemic diseases in Traditional Chinese Medicine (TCM) constitute an essential part of Chinese medical science. TCM has accumulated rich theoretical and practical experiences in the prevention and treatment of epidemic diseases, forming the academic system of epidemic febrile disease, providing robust support for epidemic prevention and resistance in TCM. However, the numerous and complex literature on TCM epidemic diseases brings challenges to the organization and discovery of epidemic disease knowledges. Objective: To leverage the powerful knowledge learning ability of state-of-the-art LLMs (LLMs) to address the efficient acquisition and utilization of TCM epidemic disease knowledges. Methods: By collecting content related to epidemic diseases from 194 ancient TCM books, as well as the knowledge graph of TCM epidemic disease prevention and treatment, we built the large TCM epidemic disease model EpidemicCHAT based on the ChatGLM model. To assess the performances of the model, several open-source LLMs were compared in the study. Results: Compared to traditional LLMs, which may fail to answer or produce hallucinations in the field of TCM epidemic diseases, EpidemicCHAT demonstrates superior answering and reasoning abilities. In the evaluation of TCM epidemic disease prescription generation, the model achieved scores of 44.02, 61.10, and 59.40 on the BLEU-4, ROUGE-L, and METEOR metrics, respectively. Conclusion: The EpidemicCHAT model proposed in this study performs excellently in the field of TCM epidemic diseases, which might provide a reference for the construction of TCM LLMs and applications such as TCM auxiliary diagnosis and Chinese herbal prescription generation.

As an acute disease, facial paralysis has high requirements for the point of treatment, and if we cannot accurately grasp the period in which the patient is suffering from the disease, we may miss the optimal time for treatment, which may affect the condition. We need to establish intelligent auxiliary diagnosis methods for facial palsy, in which AI-based methods rely on a large number of training samples. However, facial paralysis image samples are small, resulting in low model accuracy, which cannot meet the needs of clinical applications. Therefore, we plan to expand the facial palsy samples through data augmentation methods to reduce the training sample requirements and improve the accuracy of facial palsy grading. We propose the method of fusing generative adversarial networks, where the collected facial paralysis samples are trained by DragGAN model to expand the sample space, and then input into ResNet152 model for classification training. The classification effect of facial palsy images is effectively improved by the principle of data augmentation before classification. The accuracy of this experimental method on the MEEI facial palsy dataset is 85.5%, the recall is 81.4%, and the F1-score is 80.7%, which are 13.1%, 16.7%, and 15.8%, respectively compared to the method without data augmentation. A facial palsy classification model incorporating generative adversarial networks achieves higher classification accuracy compared to methods without data augmentation, providing a new approach to facial palsy classification.

With the influence of various comprehensive factors such as people’s diet and living habits and the external environment, lung cancer has become a high-incidence and high-risk disease in the world. In order to prevent and treat lung cancer, a lung cancer question and answer system based on knowledge graph is built to provide intelligent auxiliary diagnosis and treatment, which can quickly and accurately answer the questions raised by patients. In this article, the lung cancer medical case data is used to construct the knowledge graph of lung cancer, and the problem data set is constructed from the template and enhanced to increase the data volume and diversity of the data. Then build a multi-task learning model, and perform question sentence intent recognition and question sentence entity recognition at the same time, which can learn the relationship between tasks, improve the effect of the two types of tasks, and shorten the training and inference time, which can effectively reduce training cost. Finally, by analyzing the questions through the model, the entity and relationship are obtained, and the answer is obtained by querying from the knowledge graph.

In order to stablish and evaluate the drug-disease network, we take the common thyroid diseases in daily life as an example to study the modeling and evaluation of drug-disease network. Firstly, mining the highest-frequency drug in the medical records of thyroid tumors of overseas medical masters; Then, find out the thyroid tumor target and drug target in the disease target database and the traditional Chinese medicine pharmacology analysis platform, and compare and find out the potential active target of the drug on thyroid tumor. After signal pathway analysis, chemical composition target network and target signal pathway network model diagrams were constructed. The experimental results show that there are active components in the drug pair, which act on thyroid tumors through multiple signal pathways. The establishment and evaluation of drug disease network can provide a theoretical basis for the inheritance of clinical experience of famous traditional Chinese medicine.

Objective: Designs a XGBoost-based data quality control model for the traditional chinese medicine (TCM) inpatient medical record home page. Exploring the method of data normalization on the TCM inpatient medical record home page. Methods: Taking the data on the TCM inpatient medical record home page of a hospital in Jiangsu Province as the original data. Using correlation analysis to filter out some data items that have a higher degree of correlation with the data items to be quality control. Establishing a XGBoost-based data quality control model for the TCM inpatient medical record home page. Using the hierarchical 10-fold cross-validation to evaluate the model. Results: The experimental results show that the accuracy rate of the model can reach 88.60%. Conclusion: The data quality control model on the TCM inpatient medical record home page is conducive to improving the quality of the data on the TCM inpatient medical record home page and provides data support for medical research.

Objective. To investigate the medication rules and mechanism of Chinese medicine for lung cancer treated by professor Zhou Zhongying. Methods. Medical records of lung cancer patients treated by Professor Zhou were obtained and analyzed by data mining methods to arrive at the core herb combination. The relevant techniques of network pharmacology were applied to obtain the effective compounds, targets and signal pathways of the herb combination. Results. A total of 457 prescriptions were selected, involving 279 herbs. The high-frequency core herbs for lung cancer treatment were Glehniae Radix (Beishashen), Cremastrae Pseudobulbus (Shancigu), Ophiopogonis Radix (Maidong), Adenophprae Radix (Nanshashen). Through association rules and cluster analysis, the core prescriptions of 4 herbs were obtained. The four herbs were screened to obtain 19 active ingredients, and 193 potential targets were related to lung cancer treatment. Core herb combination perform its function for lung cancer by regulating the targets, such as AKT1, VEGFA, IL6, MAPK3, HIV-1A, TP53.The GO analysis results showed that a total of 285 GO entries were obtained, involving inflammatory response, and apoptotic process. The KEGG analysis was involved 113 terms, which showed that the targets of core prescription for lung cancer treatment mainly focused on PI3K-Akt, MAPK, MicroRNAs in cancer, HIF-1signaling pathway, and HTLV-1 infection signaling pathways. Conclusion. This study obtained the core prescription of professor Zhou in treatment of lung cancer by data mining techniques and explored the action mechanisms of the core herb combination by network pharmacology. This article reveals the core herb combination make effects through multicomponent, multi-target, and multi-pathway, and provides reference for clinical related research.

Objective: To study the extraction method of structured information, and realize the structuration and standardization of inpatient electronic medical records of TCM (Traditional Chinese Medicine). Methods: Based on the key terms included in "WS 445-2014 Electronic Medical Records Basic Data Set", a key set of electronic medical records for TCM hospitalization was constructed. Secondly, based on the string similarity algorithm and entity matching technology, the diagnosis and treatment information in the electronic medical records of TCM is extracted, and key-value pairs were formed with the keyword set to establish a structured diagnosis and treatment database. Finally, the diagnosis of traditional Chinese and Western medicine, physical and chemical examination and prescription data were further standardized to build a standardized diagnosis and treatment database. Results: The experimental results showed that the KS-CCD(Keyword sequencing of Clinical case data) method proposed in this paper can effectively extract information from inpatient electronic medical records of TCM. Conclusion: KS-CCD method is suitable for the extraction of structured information of electronic medical records of TCM. It provides rich research data for scientific research, and is conducive to the inheritance and development of experience of TCM.

Traditional Chinese medicine is a treasure house of the Chinese nation. Using data mining method to study traditional Chinese medicine can summarize the experience of traditional Chinese medicine and promote the development of traditional Chinese medicine. The dose-effect relationship of traditional Chinese medicine mainly refers to that the effect of drugs changes with the dose of drugs in a certain range, and there is a great relationship between the effect produced by different combinations of drugs and the dose of drugs. With the development of data mining technology, it is widely used to study the dose-response relationship between drugs with the data of famous old traditional Chinese medicine doctor's medical record. In this paper, based on the characteristics of traditional Chinese doctor's medicine data, the Apriori algorithm is improved from the construction of the data model, so that the transaction database does not need to be converted into Boolean database in the mining, which can improve the reseracher's friendliness. The results of the excavation not only included the common drugs used by famous tra-dional Chinese medicine doctors to treat lung cancer, but also analyzed the proportional relationship between the dosage of different drugs. To provide a reference for clinical researchers in the actual use of drugs between the proportion of the relationship.

Objective: To propose a method for clinical terminology extraction of traditional Chinese medicine (TCM). Methods: Conditional probability was used to measure the internal aggregation of terminologies while information entropy measuring the boundary degree of freedom (DOF), so as to establish a method to extract the clinical terminologies of TCM. Different model thresholds were set up to analyze the medical cases with 29588 clinic visits. Results: When he maximum length of character was 5, a total of 7237 terminologies were extracted, in which 5963 were correct, with precision rate, recall rate, and F value of 82.40%, 77.46% and 79.85%, respectively, and the model efficacy was the optimum. Conclusion: Internal aggregation combined with boundary DOF of characters can effectively extract clinical terminologies of TCM. In subsequent study, multiple methods should be combined to extract the terminologies in different parts of medical cases, so as to increase the precision of terminology extraction.

Purpose: To study and establish the platform of information collection and analysis of the first page of medical records in the key Medical College of TCM(Traditional Chinese Medicine). Method: According to the formulated by the State Administration of traditional Chinese medicine, the Part of the Project Filling Explanation of Inpatient Medical Record Home Page, the key indicator system for the acquisition of the first page of medical record, verification data, as well as system platform function system, the overall architecture and technology implementation were determined. Results: As the unified platform for the collection, analysis and standardization management in common use of the first page of the medical record for state, provinces (municipalities and autonomous regions), administration of traditional Chinese medicine, key medical college of traditional Chinese Medicine, was used practically in the 578 units, and was verified the feasibility of the design platform. Conclusion: Through the practical application of medical units in various provinces and cities, the platform is proved to be ease of operation and feasibility, which provides a basic platform for the construction of traditional Chinese medicine.

Objective: To study the distribution rules of traditional Chinese medicine (TCM) diagnostic information in patients with CHD angina pectoris. Methods: The clinical data of 417 patients with CHD angina pectoris were collected and analyzed retrospectively. A complex network of TCM diagnostic information was set up using Liquorice and the network nodes were clustered. Results: Two layers of networks were generated. The first layer included 17 symptomatic nodes focusing on seven symptom nodes including chest distress, chest pain, palpitation, dark purple tongue, chest prickling, dizziness and string pulse. The 15 nodes in the second layer was clustered into 6 symptomatic groups as follows: (1)dysphoria with feverish sensation in chest, palms and soles, and dreaminess; (2)indentation tongue, bulgy tongue, and loose stool; (3)amnesia, aversion to cold, and white coating;(4)nausea and vomiting, and edema of lower legs; (5)chest dull pain, and light white tongue; (6)weak pulse, palpitation, and weak breath and laziness to speak. Conclusions: The main symptoms of patients with CHD angina pectoris primarily include chest distress, chest pain, palpitation and so on, or accompanied with symptoms like aversion to cold, poor spirit and fatigue, short breath, greasy coating, yellow coating, slippery pulse, pant, insomnia and sweating, etc. This disease is located in heart, and is in close association with lung, spleen, liver and kidney. It is primarily characterized by blood stasis, accompanied with yang deficiency, blood deficiency, qi deficiency, as well as phlegm and dampness.

A novel quantization watermarking method is presented in this paper, which is developed following the established feature modulation watermarking model. In this method, a feature signal is obtained by computing the normalized correlation (NC) between the host signal and a random signal. Information modulation is carried out on the generated NC by selecting a codeword from the codebook associated with the embedded information. In a simple case, the structured codebooks are designed using uniform quantizers for modulation. The watermarked signal is produced to provide the modulated NC in the sense of minimizing the embedding distortion. The performance of the NC-based quantization modulation (NCQM) is analytically investigated, in terms of the embedding distortion and the decoding error probability in the presence of valumetric scaling and additive noise attacks. Numerical simulations on artificial signals confirm the validity of our analyses and exhibit the performance advantage of NCQM over other modulation techniques. The proposed method is also simulated on real images by using the wavelet-based implementations, where the host signal is constructed by the detail coefficients of wavelet decomposition at the third level and transformed into the NC feature signal for the information modulation. Experimental results show that the proposed NCQM not only achieves the improved watermark imperceptibility and a higher embedding capacity in high-noise regimes, but also is more robust to a wide range of attacks, e.g., valumetric scaling, Gaussian filtering, additive noise, Gamma correction, and Gray-level transformations, as compared with the state-of-the-art watermarking methods.