Ultrasound swallowing assessment: the impact of image post-processing on measurement reliability

The use of ultrasound for swallowing assessment is promising; yet its application has not translated from research into routine clinical practice. While sufficient reliability is key to support implementation, insufficient reliability for swallowing measures has been documented for pocket-sized ultrasound technology. Image quality, a known prerequisite for reliability, may be compromised in small-sized ultrasound technology. This study explored the impact of ultrasound image post-processing on measurement reliability of swallowing parameters. Ultrasound images (n = 30) of hyoid excursion, thyrohyoid approximation, and submental muscles were selected from a database of images previously collected using a pocket-sized ultrasound system in healthy adults. For this pilot study, images were measured at two time points by four raters with and without experience in ultrasound imaging. For those with experience, their engagement with ultrasound was gained in the context of research data collection, rather than daily clinical practice. At both time points, raters measured each image once with and once without the application of an image processing software that was developed to improve image contrast and image features. ICCs were calculated using linear mixed effects analyses to evaluate inter- and intra-rater measurement reliability. Intra- and inter-rater ICCs were similar for processed and unprocessed images for most parameters of interest with a few ICCs being higher or lower for processed images compared to unprocessed images. There is no strong evidence to suggest general beneficial effects of post-processing on ultrasound swallowing measurement reliability. These preliminary pilot data provide a foundation and directions for further research to clarify the role of ultrasound image post-processing for swallowing evaluation. Further refinement of processing methods may produce improved results.