{"title":"Application of deep learning to pressure injury staging.","authors":"Han Liu, Juan Hu, Jieying Zhou, Rong Yu","doi":"10.12968/jowc.2024.33.5.368","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>Accurate assessment of pressure injuries (PIs) is necessary for a good outcome. Junior and non-specialist nurses have less experience with PIs and lack clinical practice, and so have difficulty staging them accurately. In this work, a deep learning-based system for PI staging and tissue classification is proposed to help improve its accuracy and efficiency in clinical practice, and save healthcare costs.</p><p><strong>Method: </strong>A total of 1610 cases of PI and their corresponding photographs were collected from clinical practice, and each sample was accurately staged and the tissues labelled by experts for training a Mask Region-based Convolutional Neural Network (Mask R-CNN, Facebook Artificial Intelligence Research, Meta, US) object detection and instance segmentation network. A recognition system was set up to automatically stage and classify the tissues of the remotely uploaded PI photographs.</p><p><strong>Results: </strong>On a test set of 100 samples, the average precision of this model for stage recognition reached 0.603, which exceeded that of the medical personnel involved in the comparative evaluation, including an enterostomal therapist.</p><p><strong>Conclusion: </strong>In this study, the deep learning-based PI staging system achieved the evaluation performance of a nurse with professional training in wound care. This low-cost system could help overcome the difficulty of identifying PIs by junior and non-specialist nurses, and provide valuable auxiliary clinical information.</p>","PeriodicalId":17590,"journal":{"name":"Journal of wound care","volume":null,"pages":null},"PeriodicalIF":1.5000,"publicationDate":"2024-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of wound care","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.12968/jowc.2024.33.5.368","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"DERMATOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
Objective: Accurate assessment of pressure injuries (PIs) is necessary for a good outcome. Junior and non-specialist nurses have less experience with PIs and lack clinical practice, and so have difficulty staging them accurately. In this work, a deep learning-based system for PI staging and tissue classification is proposed to help improve its accuracy and efficiency in clinical practice, and save healthcare costs.
Method: A total of 1610 cases of PI and their corresponding photographs were collected from clinical practice, and each sample was accurately staged and the tissues labelled by experts for training a Mask Region-based Convolutional Neural Network (Mask R-CNN, Facebook Artificial Intelligence Research, Meta, US) object detection and instance segmentation network. A recognition system was set up to automatically stage and classify the tissues of the remotely uploaded PI photographs.
Results: On a test set of 100 samples, the average precision of this model for stage recognition reached 0.603, which exceeded that of the medical personnel involved in the comparative evaluation, including an enterostomal therapist.
Conclusion: In this study, the deep learning-based PI staging system achieved the evaluation performance of a nurse with professional training in wound care. This low-cost system could help overcome the difficulty of identifying PIs by junior and non-specialist nurses, and provide valuable auxiliary clinical information.
目的：准确评估压力性损伤（PIs）是获得良好疗效的必要条件。初级护士和非专科护士对压伤的经验较少，缺乏临床实践，因此难以对压伤进行准确分期。在这项工作中，提出了一种基于深度学习的 PI 分期和组织分类系统，以帮助提高其在临床实践中的准确性和效率，并节约医疗成本：方法：从临床实践中收集了 1610 例 PI 及其相应的照片，由专家对每个样本进行准确分期和组织标记，训练基于掩膜区域的卷积神经网络（Mask R-CNN, Facebook Artificial Intelligence Research, Meta, US）对象检测和实例分割网络。建立了一个识别系统，对远程上传的 PI 照片的组织进行自动分期和分类：在 100 个样本的测试集上，该模型的阶段识别平均精度达到了 0.603，超过了参与对比评估的医务人员（包括一名肠胃治疗师）的识别精度：在这项研究中，基于深度学习的 PI 分期系统达到了受过伤口护理专业培训的护士的评估水平。这一低成本系统有助于克服初级和非专业护士识别 PI 的困难，并提供有价值的辅助临床信息。
期刊介绍：
Journal of Wound Care (JWC) is the definitive wound-care journal and the leading source of up-to-date research and clinical information on everything related to tissue viability. The journal was first launched in 1992 and aimed at catering to the needs of the multidisciplinary team. Published monthly, the journal’s international audience includes nurses, doctors and researchers specialising in wound management and tissue viability, as well as generalists wishing to enhance their practice.
In addition to cutting edge and state-of-the-art research and practice articles, JWC also covers topics related to wound-care management, education and novel therapies, as well as JWC cases supplements, a supplement dedicated solely to case reports and case series in wound care. All articles are rigorously peer-reviewed by a panel of international experts, comprised of clinicians, nurses and researchers.
Specifically, JWC publishes:
High quality evidence on all aspects of wound care, including leg ulcers, pressure ulcers, the diabetic foot, burns, surgical wounds, wound infection and more
The latest developments and innovations in wound care through both preclinical and preliminary clinical trials of potential new treatments worldwide
In-depth prospective studies of new treatment applications, as well as high-level research evidence on existing treatments
Clinical case studies providing information on how to deal with complex wounds
Comprehensive literature reviews on current concepts and practice, including cost-effectiveness
Updates on the activities of wound care societies around the world.
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
