Automatic Posture and Limb Detection for Pressure Ulcer Risk Assessment

Pressure ulcers' problem is one of the most acute and costly health conditions. It causes suffering, pain, and extensive treatment. Furthermore, the treatment of pressure ulcers puts a huge burden on health care providers. In this paper, a system capable of assessing the risk of ulcer of each at-risk limb is proposed using the publicly available data acquired from a commercial pressure Mat. The system first classifies sleeping postures and then identifies limbs and finally tracks and profiles the impact of the applied pressure on each limb. Based on these profiles, the system generates an alert before pressure ulcer is developed. This system reduces the total expense of nursing staff and their role in patient repositioning while also lowering the risk of pressure ulcer development. A significant amount of experimental work has been conducted in order to choose suitable algorithms and parameters for each stage of the system: starting from the pressure image preprocessing, then feature extraction and posture classification, and ending by limb identification and profiling stage. Satisfactory and promising results have been obtained; where sleep posture classification achieved 99.6% overall accuracy using feed-forward artificial neural network. The limb identification is performed using advanced clustering techniques and evaluated using the silhouette index, which shows a high-quality index. Predicting the risk of pressure ulcer formation for each at-risk region is reported graphically. It was evident from the results that the proposed system would contribute significantly to the enhancement of the human body lying posture monitoring systems.