Lake to use machine learning to better understand elbow injury
Spencer Lake's lab will study elbow injuries from the cellular level using noninvasive imaging
Traumatic injury to the elbow, such as a joint dislocation or fracture from a fall, leads to long-term loss of motion, contracture and osteoarthritis in about half of all of those injured, but researchers are not sure why. In addition, treatments to prevent or relieve the problems often fail to return the elbow to pre-injury motion and function.
In addition, clinicians often struggle to predict disease stage and treatment prognosis due to limited cellular resolution in noninvasive imaging. As a result, there is a need to unravel the injury-induced cellular biology to develop new treatment strategies and diagnostic technologies, a challenge researchers at Washington University in St. Louis seek to tackle.
Spencer Lake, associate professor of mechanical engineering & materials science, has received a one-year, $40,000 grant from the Musculoskeletal Research Center at Washington University School of Medicine to link cellular resolution detail with noninvasive imaging using machine-learning approaches. Lake’s group, which developed the first elbow-specific animal injury model, plans first to develop machine learning algorithms to automatically evaluate the cells and tissues within a rat elbow at the histological level, or cellular resolution. Next, the machine-learning-derived histological detail will be used as classifiers to develop and subsequently link other machine-learning algorithms specific to noninvasive imaging parameters of the same tissues. Ultimately, Lake’s group expects that machine-learning algorithms can predict histology’s cellular-level detail based on the input of noninvasive imaging parameters, which will predict the disease state of the elbow.
Michael David, a postdoctoral research fellow in Lake’s lab, has done preliminary work on this project, and will be joined by Ulugbek Kamilov, assistant professor of computer science & engineering and of electrical & systems engineering; Aaron Chamberlain, MD, associate professor of orthopaedic surgery and a shoulder and elbow surgeon at Washington University School of Medicine; and Necat Havlioglu, MD, a pathologist at the VA St. Louis Health Care System.
The study will be the first to state-of-the-art machine learning with histology and noninvasive imaging directly. The results will help determine the feasibility of such a machine-learning approach and accelerate the understanding of elbow injuries to develop effective treatments for these and other orthopedic injuries.
