在临床实践中利用图像处理技术进行伤口管理与评估：确立在常规伤口护理中实施人工智能的可行性。

Utilizing Image Processing Techniques for Wound Management and Evaluation in Clinical Practice: Establishing the Feasibility of Implementing Artificial Intelligence in Routine Wound Care.

作者信息

Dabas Mai, Kapp Suzanne, Gefen Amit

机构信息

Mai Dabas is Master's Degree Student, Department of Biomedical Engineering, Faculty of Engineering, Tel Aviv University, Tel Aviv, Israel. Suzanne Kapp, PhD, RN, is Clinical Associate Professor, School of Health Sciences, Faculty of Medicine, Dentistry and Health Sciences, Department of Nursing, The University of Melbourne, Melbourne, Australia; and National Manager Wound Prevention and Management, Regis Aged Care, Camberwell, Victoria, Australia. Amit Gefen, PhD, is Professor of Biomedical Engineering and the Herbert J. Berman Chair in Vascular Bioengineering, Department of Biomedical Engineering, Faculty of Engineering, Tel Aviv University, Tel Aviv, Israel; Skin Integrity Research Group (SKINT), University Centre for Nursing and Midwifery, Department of Public Health and Primary Care, Ghent University, Ghent, Belgium; and Department of Mathematics and Statistics and the Data Science Institute, Faculty of Sciences, Hasselt University, Hasselt, Belgium. Acknowledgments: This work was supported by a competitive grant from the Victorian Medical Research Acceleration Fund, with funding co-contribution from the Department of Nursing at the University of Melbourne, the Melbourne Academic Centre for Health, and Mölnlycke Health Care. This work was also partially supported by the Israeli Ministry of Science & Technology (Medical Devices Program grant no. 3-17421, awarded to Professor Amit Gefen in 2020). The authors thank Ms Carla Bondini for her assistance with data collection and management for this study and Mr Daniel Kapp for proofreading the manuscript. The authors have disclosed no other financial relationships related to this article. Submitted February 1, 2024; accepted in revised form April 16, 2024.

出版信息

Adv Skin Wound Care. 2025;38(1):31-39. doi: 10.1097/ASW.0000000000000246.

DOI:10.1097/ASW.0000000000000246

PMID:39836554

Abstract

OBJECTIVE

To develop a generalizable and accurate method for automatically analyzing wound images captured in clinical practice and extracting key wound characteristics such as surface area measurement.

METHODS

The authors used image processing techniques to create a robust algorithm for segmenting pressure injuries from digital images captured by nurses during clinical practice. The algorithm also measured the real-world wound surface area. They used the hue-saturation-value color space to analyze red color values and to detect and segment the wound region within the entire image. To assess the accuracy of the algorithm's wound segmentation, the authors compared the results against wound image annotations.

RESULTS

The algorithm performed impressively, achieving an intersection-over-union score of up to 0.85 and 100% intersection with the annotations. The algorithm effectively analyzed wound images obtained during clinical practice and accurately extracted the surface area of the documented pressure injuries. These results support the feasibility and applicability of this methodology.

CONCLUSIONS

Accurate determination of wound size and healing supports decision-making regarding treatment and is essential to successful outcomes. This innovative approach for visual assessment of chronic wounds highlights the potential of computerized wound analysis in clinical practice. By leveraging advanced computational techniques, healthcare providers can gain valuable insights into wound progression, enabling more accurate assessments to support their decision-making.

摘要