Constructing Medical Image Domain Ontology with Anatomical Knowledge

Medical image reporting is an important information carrier for medical staff to record medical activities for patients. It contains a large number of technical terms and medical knowledge. Extracting effective information from medical imaging reports can better serve clinical decision making and promote the development of the medical field. This paper focuses on the breast medical imaging report, analyzes the structural characteristics of the report, designs the medical record structured template, extracts the text features from the image report, and forms the structured data in the canonical form. In this paper, the machine learning model bidirectional CNNs-LSTM-CRF is used to extract the characteristic information of related lesions in the image report. The experimental data comes from an imaging inspection report provided by a medical institution to evaluate the effect of the structure by predicting the BI-RADS classification information. The data was provided by a medical institution, and the extracting result of the feature tagging is that the average accuracy is 95.71%, the average recall is 98.10%, and the average F1 values 97.29%.

Rethinking masked image modelling for medical image representation

This article proposes a novel multitask learning framework that integrates contrastive learning and natural language generation (NLG) to enhance medical image classification and report generation. The goal is to improve disease classification accuracy and interpretability in medical diagnostics. The model architecture consists of a Vision Transformer (ViT) as a visual encoder, a transformer-based text encoder, and a multimodal decoder. The visual encoder processes medical images, while the text encoder handles disease-related text prompts. These components are trained jointly using image-text contrastive loss and language generation loss. Evaluations on the MIMICCXR and Chexpert datasets show that the model with NLG (Plain + NLG) outperforms the baseline contrastive learning model (Plain) in disease classification. For example, in the MIMICCXR dataset, the accuracy for Atelectasis increased from 17.44%(Plain) to 41.5% (Plain + NLG), and for Cardiomegaly, it improved from 19.25% to 47.4%. In Chexpert, the accuracy for Atelectasis increased from 12.5% to 58.5%, and for Pleural Effusion, from 61.10% to 64.0%. The model also demonstrated improvements in F1 scores, particularly for complex diseases like Cardiomegaly and Consolidation. The proposed multitask framework effectively combines contrastive learning with NLG, leading to improved disease classification and medical report generation. This approach has potential clinical applications by enhancing AI’s interpretability and accuracy in medical decision-making.

The occurrence of geohazards entails sudden, unpredictable, and cascading effects, with numerous conceptual frameworks and intricate spatiotemporal relationships existing between hazard events. Presently, the absence of a unified mechanism for describing and expressing geohazard knowledge poses substantial challenges in terms of sharing and reusing domain-specific knowledge pertaining to geohazards. Therefore, it is imperative to address the issue of constructing a cohesive descriptive model that facilitates the sharing and reuse of geohazard knowledge. In this study, we propose a multilayered ontology construction method tailored specifically for the domain of landslide geological hazards. By comparing existing methods, we establish a hierarchical structure and expression framework for the geological hazard ontology. Notably, our approach seamlessly integrates the conceptual and semantic layers in the relationship description at each level, enabling association representation of hazard data across multiple tiers. We define essential concepts and attributes related to landslide geological hazards, along with their respective interrelationships. To achieve effective knowledge sharing and reuse, we model the ontology of the landslide geological disaster domain using the Web Ontology Language (OWL). This modeling approach serves as a powerful tool that facilitates the sharing and reuse of disaster-related knowledge. Finally, we verify the method's validity and reliability by employing illustrative case studies. The results demonstrate that the proposed approach imposes an affordable workload on human resources. Additionally, the foundational domain ontology significantly enhances information retrieval performance, thereby yielding satisfactory outcomes.

In September 2006, the Royal Australian and New Zealand College of Radiologists (RANZCR) endorsed the modified Peer Review Audit Tool (PRAT). We aimed to assess the feasibility of using this tool in a busy radiation oncology department using an electronic medical record (EMR) system, identify areas of compliance and assess the impact of the audit process on patient management. Fortnightly random clinical audit was undertaken by using the revised RANZCR PRAT in the departments of radiation oncology at Liverpool and Macarthur Cancer Therapy Centres (LCTC and MCTC). Following audit of the EMR, treatment plans were audited by peer review. Data were collected prospectively from June 2007 to June 2008. Audits were carried out on 208 patients. Behaviour criteria were well documented in the EMR, but scanning of histology and medical imaging reports did not occur in up to a third of cases. With electronic prescriptions, treatment prescription errors were rare. In total, 8 (3.8%) out of 208 patients had a change to management recommended. Variability in interpretation of PRAT 'protocol/study' criteria was identified. We found that real-time audit is feasible and effective in detecting both issues with documentation in the EMR, and a small number of patients in whom a change to management is recommended. Recommendations have been made in order to continue to improve the audit process including documentation of any changes recommended and whether the recommended change occurred.

The automatic description of images has made good advances, but limited by the little-sample dataset, that the automatic generation of medical imaging reports is still a challenging problem. Aiming at the problem of training the joint model (CNN-RNN) on little-sample datasets, this paper proposes an improved encoding-decoding mode, in which the encoder uses less parameter in FCN (Fully Convolutional Network) for identifying lesions in mammography, and encoding it into a semantic vector. The decoder uses a LSTM (Long Short-Term Memory network) for solving, thereby reducing sample requirements. In addition, this paper combines multi-label classification (MLC) to assist the semantic coding process and uses post-processing such as the beam search to make the output fit in the natural language description better. Compared to existing models, our improved model on public mammography dataset (INbreast) with real-world data supplement achieved the BLEU score improvements by two points.

Domain ontology, encompassing both concepts and entities, along with their relations and properties, is a new medium for the storage and propagation of domain specific knowledge. For all of its associated advantages, a great deal of effort must still be expended during ontology construction, including collecting the domain-related vocabularies, developing the domain concept hierarchy, defining the properties of each concept and relationships between concepts. In recent years, several engineering handbooks have described detailed domain knowledge via organization into categories, sections, and chapters with indices in the appendix. In this paper, an approach is proposed with a view to extracting concepts, entities, properties and relationships from a handbook of a specific domain. The extracted information can then be reorganized and converted into OWL (Web Ontology Language) format to represent the domain ontology. The generation of an earthquake engineering ontology from an Earthquake Engineering Handbook is used to illustrate the proposed

Domain ontology, which is widely used in knowledge engineering, artificial intelligence and semantic Web domain, describes concepts of the entities and mutual relationships between them. However, there is no existing research on construction of database domain ontology. Domain ontology of database will help to optimize associative knowledge learning in teaching and develop the intelligent tutoring system. It can also maximize domain knowledge sharing and reusing. In this paper, database domain ontology is constructed, and then we achieve the visualization. Finally we also implement the persistent storage of ontology and query.

BackgroundDiabetic retinopathy (DR) is a leading cause of blindness in American adults. If detected, DR can be treated to prevent further damage causing blindness. There is an increasing interest in developing artificial intelligence (AI) technologies to help detect DR using electronic health records. The lesion-related information documented in fundus image reports is a valuable resource that could help diagnoses of DR in clinical decision support systems. However, most studies for AI-based DR diagnoses are mainly based on medical images; there is limited studies to explore the lesion-related information captured in the free text image reports.MethodsIn this study, we examined two state-of-the-art transformer-based natural language processing (NLP) models, including BERT and RoBERTa, compared them with a recurrent neural network implemented using Long short-term memory (LSTM) to extract DR-related concepts from clinical narratives. We identified four different categories of DR-related clinical concepts including lesions, eye parts, laterality, and severity, developed annotation guidelines, annotated a DR-corpus of 536 image reports, and developed transformer-based NLP models for clinical concept extraction and relation extraction. We also examined the relation extraction under two settings including ‘gold-standard’ setting—where gold-standard concepts were used–and end-to-end setting.ResultsFor concept extraction, the BERT model pretrained with the MIMIC III dataset achieve the best performance (0.9503 and 0.9645 for strict/lenient evaluation). For relation extraction, BERT model pretrained using general English text achieved the best strict/lenient F1-score of 0.9316. The end-to-end system, BERT_general_e2e, achieved the best strict/lenient F1-score of 0.8578 and 0.8881, respectively. Another end-to-end system based on the RoBERTa architecture, RoBERTa_general_e2e, also achieved the same performance as BERT_general_e2e in strict scores.ConclusionsThis study demonstrated the efficiency of transformer-based NLP models for clinical concept extraction and relation extraction. Our results show that it’s necessary to pretrain transformer models using clinical text to optimize the performance for clinical concept extraction. Whereas, for relation extraction, transformers pretrained using general English text perform better.

In recent automated and integrated manufacturing, so-called intelligence skill is becoming more and more important and its efficient transfer to next-generation engineers is one of the urgent issues. In this paper, we propose a new approach without costly OJT (on-the-job training), that is, combinational usage of a domain ontology, a rule ontology and a rule-based system. Intelligence skill can be decomposed into pieces of simple engineering rules. A rule ontology consists of these engineering rules as primitives and the semantic relations among them. A domain ontology consists of technical terms in the engineering rules and the semantic relations among them. A rule ontology helps novices get the total picture of the intelligence skill and a domain ontology helps them understand the exact meanings of the engineering rules. A rule-based system helps domain experts externalize their tacit intelligence skill to ontologies and also helps novices internalize them. As a case study, we applied our proposal to some actual job at a remote control and maintenance office of hydroelectric power stations in Tokyo Electric Power Co., Inc. We also did an evaluation experiment for this case study and the result supports our proposal.

In this paper, an innovative multi-modal deep learning model is proposed to deeply integrate heterogeneous information from medical images and clinical reports. First, for medical images, convolutional neural networks were used to extract high-dimensional features and capture key visual information such as focal details, texture and spatial distribution. Secondly, for clinical report text, a two-way long and short-term memory network combined with an attention mechanism is used for deep semantic understanding, and key statements related to the disease are accurately captured. The two features interact and integrate effectively through the designed multi-modal fusion layer to realize the joint representation learning of image and text. In the empirical study, we selected a large medical image database covering a variety of diseases, combined with corresponding clinical reports for model training and validation. The proposed multimodal deep learning model demonstrated substantial superiority in the realms of disease classification, lesion localization, and clinical description generation, as evidenced by the experimental results.

Due to the diverse nature of business products and operations in different geographic locations, petroleum industries demand more accurate and precise information and data. Ontology is playing an important role in knowledge management and sharing, both users and system can communicate with each other using a common understanding knowledge of a domain. This study proposes a context-based ontology construction method for extracting petroleum exploration domain ontology from unstructured Chinese text documents. The proposed approach includes the steps of (i) domain documents preprocessing, (ii) concept clustering based on the fuzzy c-means, (iii) context extraction, and (iv) domain ontology construction. Experimental results show the effectiveness and feasibility of the proposed approach.

It is laborious and time consuming to build domain ontology, which models a specific domain and specifies the concepts in a particular subject. In this paper, a 5W1H-based conceptual modeling framework for domain ontology is proposed, which is used to analysis domain concepts and relations from six aspects including Who, When, Where, What, Why and How. The definition of domain ontology, domain concept and domain relation are also presented. According to the framework, the conceptual model of Science & Technology Project Ontology (STPO) in science and technology domain is designed, in which main concepts and relations within the STPO are clearly described. From the analysis, the 5W1H conceptual modeling framework can be mapped to the class model in Object-Oriented method, which is used to model things in real world. Application shows that the framework is effective to model domain knowledge and scalable to business changes and user requirements.

Using the techniques and theories of natural language processing (NLP), this paper puts forward threads and ways to build automatically domain ontologies based on text content. Through restructuring and utilizing the generally-acknowledged domain knowledge in medical domain, this paper also constructs a concept system of the multidimensional model of modern medicine towards clinic medicine, realizing the automatic construction and acquisition of modern medical knowledge specification system. This further provides theoretical foundation and technical support for automatic construction of professional domain ontologies.

Ontology is the basis of sharing and reusing knowledge on the Semantic Web, and ontology-based semantic retrieval is a hotspot of current research. Fuzzy ontology is an extension of domain ontology for solving the uncertainty problems. To represent fuzzy knowledge more effectively, this paper presents a new series of fuzzy ontology models that consists of fuzzy domain ontology and fuzzy linguistic variable ontologies, considering semantic relationships of concepts, including set relation, order relation, equivalence relation and semantic association relation etc. The process to construct linguistic variables ontology is discussed. Using ontology and RDFS, the knowledge model for product information is created. To achieve semantic retrieval, the semantic query expansion in SeRQL is constructed by semantic relations between fuzzy concepts. The application shows that these models can overcome the localization of other fuzzy ontology models, and this research facilitates the fuzzy knowledge sharing and semantic retrieval on the Semantic Web.