When it comes to muscle, tendon, and bone injuries, early diagnosis can save you from a world of hurt and lengthy rehabilitation. Researchers at Washington University in St Louis have developed algorithms that may one day – after some refinement in imaging techniques – identify tiny strains before they turn into serious injuries.
The researchers combined mechanical engineering with image-analysis techniques to create the algorithms, one of which they measured as being 1,000 times more accurate than older methods at quantifying very large stretches near tiny cracks and tears. Another algorithm predicts where cracks are likely to form.
"This extra accuracy is critical for quantifying large strains," explains Guy Genin, the co-senior investigator on the study. "Commercial algorithms that estimate strain often are much less sensitive, and they are prone to detecting noise that can arise from the algorithm itself rather than from the material being examined. The new algorithms can distinguish the noise from true regions of large strains."
More broadly, the research resulted in a number of analytical methods that can be used to quantify the magnitude of deformation in a material from a series of digital images. It's applicable not only in wound healing and diagnosis, then, but also in other fields such as structural mechanics and perhaps even in detecting glacial rifts.
"Whether it’s a bridge or a tendon, it’s vital to understand the ways that physical forces cause structures and tissues to deform so that we can identify the onset of failures and eventually predict them," says Genin, who experimented on a variety of materials in testing the algorithms, including plastic wrap, which you can see being analyzed as it stretches in the video below.
"As you pull and stretch the plastic wrap, eventually tears begin to emerge," Genin explains. "The new algorithm allowed us to find the places where the tears were beginning to form and to track them as they extended. Older algorithms are not as good at finding and tracking localized strains as the material stretches."
Tissue strain analysis is normally done with displacement field estimation, whereby before and after images are compared, followed by strain estimation, but this is prone to error when the degree of strain is high or localized. The researchers combat the problem with an algorithm they call Direct Deformation Estimation (DDE), which cuts out the intermediary step.
They also developed the Strain Inference with Measures of Probable Local Elevation (SIMPLE) algorithm, which takes the result of the DDE calculation and compares it to an established method called the Lucas-Kanade algorithm to detect tiny strains – much smaller than those detectable by existing techniques.
"We are unaware of any other techniques to detect strain concentrations as robustly or predict the onset of fracture with this precision and accuracy," the researchers wrote in a paper published in the Journal of the Royal Society Interface.
We won't see these new techniques in clinical practice just yet, as current imaging techniques, such as MRI and ultrasound, lack the required clarity and resolution. But they may already be helping senior investigator Stavros Thomopoulos in his study of why some surgeries to repair rotator cuff (the group of tendons and muscles that connects the shoulder to the upper arm) injuries ultimately fail. And with further refinement and technological improvement, the researchers believe the new algorithms may one day make it possible to predict post-operation problems before they occur.