A Computer-Aided Visual Detection and Staging Evaluation Method for Skin Pressure Injuries

For high-risk patient groups such as long-term bedridden or malnutrition, early detection of skin pressure injuries is not only related to effective wound care and healing, but also the key to preventing potential complications. In view of the problems of large differences in wound morphology, subjectivity of physician testing, and difficulty in staging evaluation of skin pressure injuries in different course of disease, this paper proposes a computer-aided visual detection and staging evaluation method for skin pressure injuries. Based on the existing dataset of skin pressure injuries, the Spatial and Channel Synergistic Attention(SCSA) is embedded in the backbone network to comprehensively capture the different wound characteristics of skin pressure injuries with the development of the disease, and the multi-scale wound features are fully integrated by combining superficial and deep features in the neck network, so as to further deepen the semantic understanding of skin pressure injuries. The experimental results show that the detection accuracy of the proposed method is 92 %, and the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{mAP}_{50}(\%)$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{mAP}_{50: 95}(\%)$</tex> indexes are improved by 14 % and 14.8 %, respectively, compared with the benchmark model, which can meet the clinical requirements of the hospital nursing department for the detection and staging evaluation of skin pressure injuries, and provide strong technical support for the management and treatment of skin pressure injuries.