Automated measurement of pressure injury through image processing

Aims and objectives. To develop an image processing algorithm to automatically measure pressure injuries using electronic pressure injury images stored in nursing documentation.& para;& para;Background. Photographing pressure injuries and storing the images in the electronic health record is standard practice in many hospitals. However, the manual measurement of pressure injury is time-consuming, challenging and subject to intra/inter-reader variability with complexities of the pressure injury and the clinical environment.& para;& para;Design. A cross-sectional algorithm development study.& para;& para;Methods. A set of 32 pressure injury images were obtained from a western Pennsylvania hospital. First, we transformed the images from an RGB (i.e. red, green and blue) colour space to a YCbCr colour space to eliminate inferences from varying light conditions and skin colours. Second, a probability map, generated by a skin colour Gaussian model, guided the pressure injury segmentation process using the Support Vector Machine classifier. Third, after segmentation, the reference ruler - included in each of the images - enabled perspective transformation and determination of pressure injury size. Finally, two nurses independently measured those 32 pressure injury images, and intra class correlation coefficient was calculated.& para;& para;Results. An image processing algorithm was developed to automatically measure the size of pressure injuries. Both inter- and intra-rater analysis achieved good level reliability.& para;& para;Conclusions. Validation of the size measurement of the pressure injury (1) demonstrates that our image processing algorithm is a reliable approach to monitoring pressure injury progress through clinical pressure injury images and (2) offers new insight to pressure injury evaluation and documentation.& para;& para;Relevance to clinical practice. Once our algorithm is further developed, clinicians can be provided with an objective, reliable and efficient computational tool for segmentation and measurement of pressure injuries. With this, clinicians will be able to more effectively monitor the healing process of pressure injuries.