Applying Machine Learning for Intelligent Assessment of Wheelchair Cushions from Pressure Mapping Images

Pressure ulcers are skin and underlying tissue injuries caused by the cells' lack of oxygen and nutrition due to blood flow obstruction from constant pressure on the skin. It is prevalent in people with motion disabilities, such as wheelchair users. For both prevention and healing, wheelchair users should occasionally change their sitting posture, use cushions that evenly distribute the pressure, or relieve pressure from the sensitive areas. Occupational therapists (OTs) often use pressure mapping systems (PMS) to assess their clients and recommend them a cushion. A cushion with more uniform pressure distribution and fewer pressure concentration points is ranked the highest. This paper offers a novel approach to enhance the objectivity of PMS readings and rankings for OTs. Our method relies on image segmentation techniques to generate quantifiable measures for cushions assessment. We implemented a sequential process to generate a score representing a cushion's suitability for an individual, which begins with PMS image segmentation using machine learning, followed by a deep learning algorithm for identifying high-risk pressure points. We introduced a Cushion Index for quantifying and ranking the cushions. Clinical Relevance- By selecting proper cushions for wheelchair users, the risk of developing PUs is reduced.