The study presents a comprehensive methodology for visual assessment of aseptic wounds with the possibility of its application in artificial intelligence (AI) systems for monitoring the wound process. The relevance of the topic is due to the growing number of surgical interventions, war wounds, and man-made injuries that are accompanied by wound formation. The main goal of the article is to develop a conceptual model of the subject area "Aseptic Wound" to support the design of intelligent medical systems, particularly those utilizing deep learning algorithms for image-based diagnosis and wound classification or segmentation. The morphological, planimetric, topographic and color markers of aseptic wounds, which can be used for automated assessment, are analyzed. A structured set of parameters is proposed, which can be integrated into annotated training datasets for machine learning models and has been used to construct a prototype ontology of aseptic wounds. This ontology enables the semantic interpretation and standardization of wound features, facilitating interoperability between clinical data sources and AI systems. The results of the work are the basis for improving the accuracy of clinical diagnosis and individualization of treatment protocols. Moreover, the conceptual model and ontology lay the groundwork for the development of an ontological knowledge base for medical expert systems, which is particularly important in the context of training deep learning models that require well-labeled, semantically rich datasets to achieve high performance and generalizability in clinical practice.
The rapid development of digital technologies and artificial intelligence has led to significant
transformations in the healthcare sector, including in the diagnosis, treatment and monitoring of
patients. In medicine, the use of AI allows for optimizing the provision of medical services, speeding
up the process of working with databases, improving the results of diagnosing visual and digital
examination results, and facilitating differential diagnosis and the diagnosis process. The use of AI
tools is especially relevant in surgery, where a quick and objective assessment of the wound
condition can determine the effectiveness of further clinical decisions. However, the issue of visual
assessment of different types of wounds and the wound process is still far from ideal and therefore
subject to detailed study and research [
Today, the number of wound defects (wounds) is growing rapidly, not only due to scientific and technological progress and, as a result, the development of man-made injuries in complex technological industries, but also, unfortunately, due to a sharp increase in the number of military
0000-0002-6559-0721 (S. Lupemko); 0000-0001-6672-3971 (M. Tomaszewski); 0000-0002-3839-459X (A. Bryniarska); 0000-0002-5506-7582 (A. Pavlyshyn); 0000-0002-8340-913X (O. Orobchuk) conflicts and wars. In addition, progress in the field of medicine contributes to an increase in the number of types of surgical interventions, changes in their quality, duration, and complexity, which in turn is accompanied by aseptic wounds, without which it is impossible to gain surgical access to tissues. An aseptic wound formed as a result of a planned surgical intervention is an ideal model for analyzing the features of the wound process. It does not contain signs of infection, and its healing occurs under favorable conditions, which allows us to identify typical visual markers of regeneration.
Healing of aseptic wounds by primary tension, without visible cosmetic defects and complications, is one of the important components of successful surgical treatment. Wound assessment is carried out according to a number of parameters that have a clear morphological, topographic, and functional classification. At the same time, digital images obtained with cameras or lidar sensors can serve as a source for building training samples in computer vision systems.
From the perspective of deep learning, the creation of a standardized, semantically rich
representation of aseptic wounds is of fundamental importance. Deep learning algorithms,
particularly convolutional neural networks (CNNs), require large, well-annotated datasets with
consistent feature definitions to effectively learn complex visual patterns [
Furthermore, the proposed ontology can be used to support explainable AI (XAI) frameworks,
enabling clinicians to understand and validate the reasoning behind the system's predictions.
According to [
In this regard, there is a need to formalize the knowledge of specialists (in particular, surgeons) and develop a standardized set of characteristics of an aseptic wound that will provide a single terminology and be suitable for automated analysis, diagnosis and monitoring - in the form of a computer ontology. The main goal of this article is to develop a conceptual model and ontology of the subject area "Aseptic Wound" that will support the design of intelligent medical systems based on deep learning. This includes the identification and formalization of key morphological, topographic, and visual parameters of aseptic wounds suitable for automated analysis. The outcome of this work serves as a foundation for constructing training datasets, improving diagnostic accuracy, personalizing treatment protocols, and enhancing the integration of ontological knowledge into medical expert systems.
Previous research in digital medicine and biomedical image analysis has focused on the use of deep
learning to identify pathological changes in the skin (e.g., melanoma, pressure ulcers, burns) [
Aseptic wounds have a complex course that depends on a number of factors - clinical, visual, and individual patient characteristics. However, techniques focused on aseptic wounds remain fragmented. There is a lack of systematized approaches to visualizing such wounds, taking into account their specific features (absence of infection, predictability of regeneration phases, controlled conditions of formation). Therefore, an approach that combines clinical relevance and technological implementation for automated assessment of aseptic wounds needs to be improved. And to formalize and structure the knowledge used by medical professionals in the visual assessment of wounds, it is advisable to create an appropriate ontology.
3.1.
The conceptual model was built on the basis of the knowledge of experts in the field of surgery through a multi-stage survey. The main stages of modeling were: 1.
Defining the goals of the ontology: providing a formalized description of key concepts, processes and signs of aseptic wounds for further use in AI systems.
Collection and systematization of concepts: based on the analysis of scientific literature and expert responses, a preliminary list of terms and attributes was formed.
Grouping of concepts into classes: all concepts are grouped into logical categories (for example: Wound type, Signs of healing, Visual characteristics, Localization, Stage of the process).
Establishment of relationships between concepts: a hierarchy of concepts was formed, as well as semantic relationships such as “part of”, “related to”, “has a feature”. 5. Formalization in the form of a diagram in the Protégé environment that reflects the conceptual structure of the ontology.
Expert knowledge was collected through a structured survey using questionnaires and semistructured interviews in the following stages (Fig. 1).
Wounds are something that is most often encountered not only in the course of their professional activities by specialists in various fields, but also by ordinary people in domestic and non-industrial settings. Let's define what we mean by a wound. A wound is a mechanical injury that results in a disruption of the integrity of the skin, mucous membranes (superficial wounds), and underlying internal organs (deep wounds). The main visual signs of a wound are the presence of damage to the skin or mucous membrane, or a combination of both.
When studying the visual characteristics of wounds, we begin with those that meet all the objective criteria of a 'wound' but exhibit minimal signs of complications such as inflammation, necrosis, or bleeding. Such wounds include the most 'healthy' ones - aseptic, i.e., surgical wounds. Aseptic (surgical) wounds are made by a surgeon under maximally sterile conditions for therapeutic or diagnostic purposes. Aseptic wounds inflicted by medical personnel using specialized surgical instruments (scalpel, surgical saw, scissors, perforators, etc.) may exhibit some visual characteristics of incised wounds: slight tissue separation in the direction of the applied force of the surgical instrument, gaping [10,11].
One of the objectives of our study is to assess the wound and the wound healing process through its visualization followed by AI training. The visualization process is based on images of the wound and surrounding tissues, which are evaluated according to selected criteria. The application of AI enables objective monitoring of the wound healing process in accordance with the wound healing phases and allows for remote adjustment of treatment strategies [12,13].
It should be noted that the visual indicators for identifying the very concept of a wound include
the presence of damage to the skin, mucous membranes, or a combination of both. These clinical
signs are effectively detected and segmented by AI-based image analysis tools [
No
Yes Subcutaneous fat
Perito- Muscle Serousneum tissue membrane Underlying internal organs
Lungs Brain
Heart
Liver
Spleen Kidneys
Intestines
Some of the key criteria for processing images of aseptic wounds and other types of wounds include determining their planimetric characteristics, namely: wound shape, dimensions, perimeter, depth (in 3D), and area. Using digital tools and AI, based on the obtained digital image, we determine the affected surface area by creating a two-dimensional or 3D (with the help of LiDAR sensors) model of the wound contours.
The criteria for planimetric wound assessment include the following indicators – table 2. The specified planimetric parameters will be used not only for aseptic wounds but also for wounds of other types, as they are universal.
Perimeter
In addition to the wound's planimetric characteristics, clear visualization and description of its edges are essential, as they serve as one of the indicators for classifying the wound according to widely accepted types. Based on the condition of the wound edges, it can be categorized as incised, puncture, lacerated, crushed, bite, gunshot, etc. Therefore, the characterization of wound edges is very important, and we propose the following visual features – table 3.
When analyzing the 2D or 3D model of the wound's visual characteristics, determining its anatomical topographical location is extremely important. Accurate wound topography on the patient’s body makes it possible to obtain a number of essential anatomical characteristics, such as: the type of underlying tissues directly at the injury site, and the location relative to vital organs and systems (major blood vessels, nerve trunks, organs, etc.). Topographical data also help effectively predict the risk of complications and plan the optimal prevention strategy.
Therefore, to optimize AI performance in the visual assessment of the wound and the overall wound healing process, we propose determining the following anatomical-topographical characteristics of the wound – table 4.
An important element of the visual assessment of a wound is its relation to Langer’s lines. These are imaginary lines that reflect the anatomical structure of the skin, running along the surface and indicating the direction of maximum tension of collagen and elastin fibers in the dermis. They correspond to the orientation of connective tissue fibers. The alignment of a wound parallel to Langer’s lines positively affects the speed and quality of healing. Such wounds heal faster and tend to form less visible scars. Conversely, wounds located perpendicular to these lines heal more slowly, experience tension at the wound edges (negatively affecting healing), and often result in wide, noticeable, and aesthetically unpleasing scars.
No
Since wound formation is often accompanied by damage to blood vessels of various types and calibers, it is essential that the visual assessment of the wound includes identification of the type of bleeding – table 6.
Each type of bleeding has its own visual characteristics that help in its identification, which in turn directly influences the wound healing process, the risk of complications, and the overall treatment strategy.
In addition to determining the visual features of the wound, it is also essential to record the time of wound formation. Documenting when the wound occurred allows the specialist to approximately determine the phase of the wound healing process and the tissue changes in response to injury. Furthermore, the longer a wound remains untreated and without primary surgical care, the more likely it is to become bacterially contaminated.
Determining the time of an aseptic wound is a simpler task, as it is inflicted by a surgeon and precisely recorded in the medical documentation down to the hour and minute. For all other types of wounds, it is recommended to document the time at least approximately. Based on the account of the injured person, witnesses, healthcare workers, or rescue personnel. The following time intervals are proposed for recording the moment of wound occurrence – table 7.
In the case of aseptic wounds, all the above-mentioned visual characteristics must of course be assessed first, especially when the analysis is performed shortly after the wound is inflicted.
However, the wound changes over time, and from the moment of tissue damage, a complex process of physiological responses is triggered in the body, aimed at quickly restoring the integrity of the injured area. This regeneration process is known as the wound healing process, and it is characterized by a specific sequence (phases) of reparative processes in the wound, surrounding tissues, and the body as a whole.
During the wound healing process, the visual characteristics of the wound and their markers will naturally evolve, and new ones may appear. Additionally, a number of complications may arise during healing, each with their own distinct visual markers.
From the moment of wound occurrence and throughout the healing process, different types of wound exudates may appear. These may come from the damaged tissues or as a result of complications such as infection. It is proposed to record wound exudates based on the following criteria – table 8.
Serous exudate
Fibrinous exudate
Obtaining reliable visual information regarding the presence or absence of wound exudates, along with their qualitative and quantitative characteristics, allows for the use of such findings as diagnostic criteria to form a clear picture of the wound healing process. Whether it is progressing successfully or, on the contrary, complications or infections are developing.
In addition to identifying the above-mentioned types of wound exudates, it is extremely important
to learn to visually diagnose the presence of a bacterial biofilm on the wound surface [
The visual indicators of bacterial biofilm on the wound surface include the following characteristics – table 9.
The presence of the above-mentioned signs on the wound surface indicates the development of complications in the healing process, including: increased inflammation, the spread of infection, the development of resistance to antibiotics and antiseptics, and, as a consequence, the transformation of the wound into a chronic state.
Visual assessment of tissue color in the wound itself and in the surrounding skin also plays a
critical role. A physiological (pale-pink) skin color is a diagnostic marker of a healthy state of the
body. Any discoloration (pallor, jaundice, cyanosis) may indicate pathological conditions such as
anemia, hypoxia, or liver dysfunction. Recent AI-based approaches have emphasized the importance
of accurate color calibration in wound image analysis. One smartphone-based system achieved
improved segmentation performance when a calibration chart was used to standardize color and
measurement across images, enabling more consistent identification of epithelialization (pink),
granulation (bright red or pale pink), and necrosis (black or white tissue) [
Based on the collected knowledge, the following options are proposed as physiological (normal) skin color characteristics – table 10.
In addition to recording the physiological skin tone, it is also necessary to document pathological skin discoloration around the wound, which is characteristic of a number of conditions that directly affect the speed and quality of wound healing and the course of the healing process.
Therefore, the following visual markers of skin tone under pathological conditions are proposed – table 11.
Depending on the development of complications, tissue death, known as necrosis, may occur during wound healing. Necrotic changes in wound tissues negatively affect healing dynamics by significantly slowing down tissue regeneration, promoting infection, and contributing to wound chronicity.
The final stage of wound healing is the scarring and epithelialization phase, during which reparative processes restore the integrity of the skin and form a scar. Successful completion of this phase is critical for restoring both the strength and integrity of the damaged tissue. The resulting scar(s) may present the following visual types of scars – table 12. Height
Area Perimeter
No
Yes Planimetric characteristics of scars
Numerical value (mm) Linear
Arcuate
Polygonal
The characteristics of the vascular pattern are also of great importance for the visual assessment
of the wound and the wound healing process itself. The vascular pattern around the wound is an
important diagnostic indicator that helps evaluate the condition of the wound and the surrounding
tissues. Studies using optical microangiography in experimental wound models have demonstrated
that dilation of collateral vessels and newly forming microvessels around the wound bed occur
rapidly during the inflammatory phase, facilitating increased perfusion to support hypoxic tissue[
Visual markers of the vascular pattern are important not only for the wound itself but also for the adjacent tissues. The following visual characteristics are proposed – table 13.
The presence of a vascular pattern indicates increased blood flow and activation of microcirculation, which is a normal physiological response to injury. However, in some cases, when combined with other clinical signs, it may be a marker of pathological processes such as infection, allergic reactions, venous stasis, microcirculation disorders, lymphangitis, sepsis, trophic disorders, etc. Therefore, the vascular pattern should be assessed only in combination with other wound markers.
A key component of successful wound healing is the formation of granulation tissue. Granulation is the process through which the body generates new connective tissue and blood vessels at the site of injury to the skin or mucous membrane. Therefore, visualization of granulation tissue is a critical part of assessing wound status. The following visual markers of granulation tissue are proposed – table 14. Hypergranulation Fibrous granulation
Pale
Pale pink
Bright pink
Red
Bluish
Gray
Black Infected granulation Necrotic granulation The set of the above tables forms the general view of the scale of visual characteristics of an aseptic wound. 3.3.
Despite the availability of a wide range of visual characteristics of wounds, the lack of a formalized model that would describe the relationships between these features greatly complicates their integration into intelligent medical systems. To overcome this problem, it is advisable to create a wound ontology - a structured representation of concepts, their attributes, types of relationships, and the hierarchy that exists between them.
The ontology allows for semantic interoperability between different data sources, medical
applications, and machine learning algorithms [
In the context of the development of intelligent diagnostic and remote monitoring systems, the ontological model of wounds is becoming a key basis for the development of knowledge-based platforms that support automated clinical decision-making. In particular, it allows you to link visual signs to clinical scenarios, healing phases, and prognosis.
As part of our study, we formed a primary prototype of the aseptic wound ontology based on the
proposed system of visual markers. This prototype describes the basic concepts, such as types of
injuries, planimetric parameters, topographic characteristics, phase signs of healing, and establishes
logical and functional relationships between them. To build the ontology of the subject area “Aseptic
Wound”, the Protégé tool environment [
The proposed methodology for assessing an aseptic wound made it possible to develop a hierarchically ordered structure of visual characteristics that covers both static morphometric features and dynamic changes in the healing process. Based on this, a basic version of the aseptic wound ontology was formed, which allows not only to formalize terminology but also to ensure interoperability between different information sources and clinical systems.
The use of an ontological model allows generalizing and unifying the process of visual description of wounds, as it links visual parameters (shape, size, type of tissue damage), phase characteristics (inflammation, proliferation, remodeling), type of bleeding, presence of a film or discharge, as well as scarring and skin color. This allows artificial intelligence systems to effectively classify and predict the course of the wound process.
Planimetric and morphological characteristics were tested to form descriptors of the visual model within the prototype ontology. The established intercategory relationships (e.g., between wound localization, planimetry, and potential complications) became the basis for building rules in logical computing modules.
Thus, the created model not only allows for standardizing the process of diagnostic description of an aseptic wound, but also ensures its machine interpretability, which is critical for integration into clinical information systems and decision support systems.
The proposed methodology has several limitations. The developed ontology is based primarily on
expert knowledge and theoretical modeling, with limited validation using real-world clinical
datasets. Similar to [
Future research should focus on expanding and refining the ontology using large-scale annotated datasets derived from real clinical images of aseptic wounds. Special attention should be given to the formal alignment of the ontology with existing biomedical standards such as SNOMED CT (Systematized Nomenclature of Medicine – Clinical Terms), FMA (Foundational Model of Anatomy), or ICD (International Classification of Diseases), to ensure semantic interoperability with electronic health records. Additionally, efforts should be made to embed the ontology within AI training pipelines, especially in convolutional neural networks and explainable AI frameworks, to evaluate its practical utility in image classification, segmentation, and prognosis prediction tasks. Further development will also involve the extension of the model to include infected, chronic, or posttraumatic wounds, thereby creating a comprehensive ontology capable of supporting intelligent wound management systems across the full spectrum of wound types.
The developed conceptual methodology for visual assessment of aseptic wounds opens up new opportunities for the integration of digital technologies into the practice of surgical diagnosis and monitoring of the wound process. Due to their predictable and uniform healing trajectory, aseptic wounds constitute an optimal reference framework for the initial training of AI models focused on wound assessment, enabling the recognition of typical regenerative features before extending model capabilities to pathological or non-standard healing scenarios.
The proposed ontological model allows formalizing the key visual characteristics of wounds, providing their unified representation for artificial intelligence systems. The use of structured descriptors and cross-category relationships helps to improve the accuracy of automated wound assessment and wound healing prediction. The results of the study can be applied both in clinical practice and to create training samples for medical AI systems, which makes the proposed approach a promising direction in the development of digital medicine.
Declaration on Generative AI
The author(s) have not employed any Generative AI tools.
M. Wierzbicki, B. Jantos, M. Tomaszewski, A Review of Approaches to Standardizing Medical Descriptions for Clinical Entity Recognition: Implications for Artificial Intelligence Implementation, Applied Sciences 14(21):99032024 (2022). doi:10.3390/app14219903. [6] M. Myslicka, et al. Review of the application of the most current sophisticated image processing methods for the skin cancer diagnostics purposes, Archives of Dermatological Research 316.4 (2024): 99. [7] Sven van Poucke and Woundontology https://www.slideshare.net/slideshow/woundontology1/424644#24.
Consortium. [8] Gene Ontology Consortium, Wound http://PMC+4informatics.jax.org+4informatics.jax.org+4. healing (GO:0042060).
URL: URL: [9] Ontology for Biomedical Investigations (OBI). URL: http://obi-ontology.org/. [10] E. Haesler, et al. A systematic review of the literature addressing asepsis in wound management, Wound Practice & Research: Journal of the Australian Wound Management Association 24.4 (2016): 208-216. [11] E. Purssell, R. Gallagher, D. Gould, Aseptic versus clean technique during wound management? Systematic review with meta-analysis, International journal of environmental health research 34.3 (2024): 1580-1591. [12] V. Borst, A. Riedmann, T. Dege, K. Müller, A. Schmieder, B. Lugrin, S. Kounev, WoundAIssist: A Patient-Centered Mobile App for AI-Assisted Wound Care With Physicians in the Loop 2506.06104 (2025). [13] H. Carrión, M. Jafari, H. Yang, R. Isseroff, M. Rolandi, M. Gomez, N. Norouzi, Healnet-selfsupervised acute wound heal-stage classification, In International Workshop on Machine Learning in Medical Imaging, Cham: Springer Nature Switzerland (2022): 446-455.
Editor and
Knowledge
Acquisition
System.
URL: [23] OWL Web Ontology Language http://www.w3.org/TR/owl-guide/.
Guide.
W3C
Recommendation.
URL: