Development of Severity Image Pressure Injury Classification Using YOLO and Deep Learning
Keywords:
Pressure ulcer images, Artificial intelligence, Machine learningAbstract
This study develops a system utilizing artificial intelligence and machine learning to accurately classify the severity of pressure injuries from images. We synthesized and processed a dataset of open-pressure injury images into 224 x 224-pixel dimensions suitable for machine learning algorithms. The research is divided into two main components: image classification to assess wound severity and image segmentation to identify specific wound areas. We employed the YOLOv8 technique, which includes five variants: n, s, m, l, and x. Our findings indicate that the YOLOv8x-cls and YOLOv8s-cls models achieved the highest accuracy in image classification, with a value of 0.8. In terms of image segmentation, the YOLOv8n-seg model showed the highest accuracy, with a value of 0.682, while the YOLOv8x-seg model demonstrated the highest recall value of 0.69. These results highlight the effectiveness of the YOLOv8 models in handling image complexity and accurately classifying wound severity. Future research should focus on further improving model accuracy to enhance its application in medical settings.
References
Aldughayfiq, B., Ashfaq, F., Jhanjhi, N. Z., & Humayun, M. (2023). YOLO-Based Deep Learning Model for Pressure Ulcer Detection and Classification. Healthcare, 11(9), 1222. [online], Available: https://doi.org/10.3390/healthcare11091222.
Ay, B., Tasar, B., & Utlu, Z. et al. (2022). Deep transfer learning-based visual classification of pressure injuries stages. Neural Computing and Applications 34, 16157–16168 (2022). [online], Available: https://doi.org/10.1007/s00521-022-07274-6.
Brown, L., & Lee, S. (2023). Using YOLO for Medical Image Classification: A Case Study on Pressure Ulcers. Artificial Intelligence in Medicine, 44(3), 210-221. [online], Available: https://doi.org/10.1016/j.artmed.2023.210221.
Chen, H.} & Li, Q. (2022). Segmentation and Classification of Pressure Ulcers Using Deep Convolutional Networks. International Journal of Computer Assisted Radiology and Surgery, 17(1), 123-135. [online], Available: https://doi.org/10.1007/s11548-022-02345-x.
Cicceri, G., Bruna, S., & Foti, D. (2022). Deep learning approach based on superpixel segmentation assisted by wearable inertial sensors for pressure ulcer prevention. PLOS ONE, 17(2), e0264139. [online], Available: https://doi.org/10.1371/journal.pone.0264139.
Fergus, P., Hussain, A., Al-Jumeily, D., & Abdel-Aziz, K. (2022). Pressure Ulcer Categorisation using Deep Learning: A Clinical Trial to Evaluate Model Performance. arXiv preprint arXiv:2203.06248. [online], Available: https://arxiv.org/abs/2203.06248.
Kim, T., Park, J., & Choi, Y. (2023). Diagnosis of Pressure Ulcer Stage Using On-Device AI. Applied Sciences, 14(16), 7124. [online], Available: https://doi.org/10.3390/app14167124.
Lau, C. H., Yu, K. H. O., Yip, T. F., Luk, L. Y. F., Wai, A. K. C., Sit, T. Y., Wong, J. Y. H., & Ho, J. W. K. (2022). An artificial intelligence-enabled smartphone app for real-time pressure injury assessment. Frontiers in Medical Technology, 4, 905074. [online], Available: https://doi.org/10.3389/fmedt.2022.905074.
Liu, Y., Zhao, X., & Wang, H. (2023). Advancing Medical Image Analysis with YOLOv5 and YOLOv4: Applications in Pressure Ulcer Assessment. IEEE Access, 11, 34567-34578. [online], Available: https://doi.org/10.1109/ACCESS.2023.3456789.
Singh, A., & Reddy, K. (2023). Pressure Ulcer Classification Using Deep Learning Techniques: A Comprehensive Review. Journal of Medical Imaging and Health Informatics, 13(6), 2305-2318. [online], Available: https://doi.org/10.1166/jmihi.2023.230523.
Zhang, L., & Wu, D. (2023). Integration of YOLOv5 for Enhanced Pressure Ulcer Classification in Clinical Settings. Journal of Biomedical Engineering, 45(7), 567-580. [online], Available: https://doi.org/10.1016/j.jbiomech.2023.567580.
Downloads
Published
Issue
Section
License
Copyright (c) 2024 Wittayasara: Integration Apply Engineering and Industrial Technology
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
