Effects of Mechanical Damage on Patellar Tendon Ultrasound Texture Parameters

Abstract

The patellar tendon assists in crucial daily motion of the knee such as knee extension. Partial tears or ruptures of this tendon have a prevalence of ~14% in elite athletes and 8.5% in nonelite athletes. Patellar tendon ruptures, caused by high demand tensile loading, often occur in tendons with pathology such as patellar tendinopathy. An effective diagnostic tool for tendon injuries is ultrasonography (US). Irregularities in US images that could indicate injury include variations in echogenicity (brightness) within the tendon structure compared to surrounding structures on the US image. Current clinical interpretation of tendon US images is generally qualitative and hence may be subjective, particularly when evaluating more subtle levels of injury. Texture analysis of US images offers an objective, quantitative approach to analyzing clinical images and can facilitate injury evaluation. Improving the connection between quantitative US parameters and diagnostics will increase the accessibility of these tools beyond the highly trained clinician for preventative use in at-risk populations such as athletes. We aim to investigate the effects of mechanically induced damage on US image-based texture parameters to explore potential imaging biomarkers of tendon injury. Juvenile porcine patellar tendons (n=13) were dissected and potted in urethane resin. Of these, 6 tendons underwent load to failure testing, where the average yield strain of 12.74% was calculated and determined to be the damage threshold. The remaining 7 samples underwent loading at 3%, 15%, and again at 3% strain, for 31 minutes each, with 20-minute relaxation (in slack) periods in between. B-mode and Shear Wave Elastography (SWE) images were collected in a preloaded, unstrained state prior to each strain hold and following the final strain hold. Images were also collected at the peak stress, midpoint (t=15 minutes), and equilibrium stress state for each applied strain magnitude. Residual strength tests were conducted following the final preload. Acquired US images were segmented, and 1st, 2nd, and higher order texture parameters were quantified using PyRadiomics. Statistical analysis was conducted to determine significant differences between texture parameters at various states of loading, as well as to assess correlations between mechanical properties and texture features. A total of 23 texture parameters (B-Mode) including various 1st, 2nd, and higher order exhibited significant differences at different loading states. Additionally, multiple texture parameters had significant correlations with various mechanical parameters. Only 1 texture parameter associated with SWE exhibited any significant differences. The results from this study indicate that 1st, 2nd, and higher order texture parameters are potentially reliable biomarkers in evaluating damage of the patellar tendon.