Virtual µCT from clinical CT image
Image-based modeling is a popular approach for performing patient-specific biomechanical simulations. Accurate modeling is critical for orthopedic applications to evaluate diagnostics, implant design, and surgical planning. It has been shown that bone strength can be estimated from bone mineral density and trabecular bone structure. However, these findings cannot be directly applied to patient-specific modeling because only bone mineral density values can be derived from calibrated clinical CT, but no information on the trabecular bone structure is available to the clinician.
Image super-resolution is a computer vision task involving the enhancement of high-resolution images from low-resolution counterparts. With the rapid development of deep learning techniques in recent years, deep learning-based super-resolution models have been actively explored and often redefine the state of the art in various benchmarks.
In this work, we proposed and implemented a method based on generative adversarial networks to predict the structure of the trabecular bone from the clinical CT scan of the patient based on a deep learning super-resolution model applied to three-dimensional datasets.
