Pixel Aspect Ratio as a Source of Variation in Quantitative Videofluoroscopy Measures.

Purpose: Pixel-based measures have become quite common in quantitative videofluoroscopy (VFSS) analysis. Recently, Larsen et al. (2023) identified non-square pixels as a potential source of error in such measures. To check for this concern, they provided instructions for measuring pixel-aspect ratio (PAR), based on the width:height ratio (W:H) of the fluoroscope tube aperture. Using this approach, we explored PAR measurement in our own data sets. Methods: From a large pool of VFSS data, 50 lateral view images were randomly selected from four different equipment setups, for a total of 200 images. Trained raters, blinded to each other’s ratings and to recording set-up, measured PAR (Larsen method) using ImageJ software. Images where the entire circulate/oval aperture was not visible, were padded to enable accurate measurement of the W:H ratio. Inter-rater reliability for PAR was calculated on a random selection of 10% of the dataset using average two-way mixed intra-class correlation coefficients for consistency (ICCs) with 95% confidence intervals (CIs). We calculated mean and standard deviation (±SD) of PAR for each of the four equipment setups. Results: Inter-rater reliability for PAR measures across the four recording systems was excellent (ICC = 0.99, CI 0.979, 0.997) and an average (±SD) absolute difference of 0.01 (± 0.005). Measures of mean PAR (±SD) across the four recording set-ups were 1.00 (±0.001), 0.99 (±0.001), 1.00 (±0.003), and 1.08 (±0.01). Upon recognition of the one setup with non-square PAR of 1.08, we were able to work with the recording equipment vendor to re-calibrate the video signal input. A check of 5 new images from that set-up confirmed correction to an average (±SD) PAR of 1.00 (±0.004).Conclusions: Pixel-based measurement of PAR using image aperture width and height were highly reliable in this dataset. The data confirms that differences across recording systems exist, and that non-squareness of pixels may be a source of measurement variation in VFSS analysis. This issue may contribute a margin of error to pixel-based measures of structural position, area and movement distance. Confirmation of square PAR is a step that should be performed at the time of installation with new fluoroscopic or recording equipment.