Background: Ultrasound has not progressed to standard clinical practice despite the fact that it offers a radiation-free and non-invasive procedure for swallowing assessment and promising validity and reliability for ultrasound measurements. The cost, which may be one of the significant obstacles to restrict integration into clinical practice, can be overcome by using new pocket-sized portable ultrasound devices. However, early research suggests that poor image quality may compromise measurement reliability. To enhance the interpretation of swallowing ultrasound images, an image enhancement software is recently developed which enhances the contrast and features of acquired ultrasound images.

Aim: The focus of the present research is to determine if this image enhancement software increases intra-and inter-rater reliability of ultrasound measurement compared to the original ultrasound images.

Method: Archived data of 40 ultrasound images were measured by the primary investigator and three co-investigators to derive intra-and inter-rater reliability. Training for co- investigators was provided one week prior to ultrasound image measurement under two conditions: with and without the image enhancement software programme. A total of 80 images were measured by each researcher, which includes 40 original images and 40 enhanced images. Trials were randomized within and across the raters. Image analysis occurred across a four-day period, with 20 trials per day to minimise the effects of fatigue. One week following the completion of inter-rater reliability measurement, all investigators re-evaluated the same images, in random sequence, to evaluate intra-rater reliability. Therefore, a total of 160 measures were made by each investigator.

Results: Intra-rater reliability for enhanced images was good (> 0.75) for hyoid excursion and moderate (0.50 – 0.75) to good (> 0.75) for floor of mouth muscles, while it ranged moderate (0.50 – 0.75) to good (> 0.75) for hyoid excursion and poor (< 0.50) to good (> 0.75) for floor of mouth muscles for normal images. For thyrohyoid approximation, intra- rater reliability was found to be similar for both enhanced and normal images, ranging from poor (< 0.50) to good (>0.75). Inter-rater reliability of enhanced images ranged from moderate (0.50 -0.75) to good (> 0.75), whereas poor (< 0.50) to moderate (0.50 – 0.75) for normal images.

Conclusion: The results from this study indicate a slight increase in reliability for enhanced images compared to normal images and this difference is not potentially large enough to be clinically significant.