There has been a large number of recently published studies describing a wide variety of promising applications of deep learning in musculoskeletal imaging. Deep learning methods have been shown to be highly efficient and accurate for segmenting musculoskeletal tissues, which may eventually allow incorporation of quantitative image analysis into clinical practice. Many studies have shown promising preliminary results for using deep learning methods for accelerating musculoskeletal MR imaging. More importantly, multiple studies have documented the feasibility of using deep learning methods for musculoskeletal disease detection with diagnostic performance comparable to human readers for identifying a wide variety of pathologic abnormalities. However, much additional technical development is needed to create fully-automated deep learning methods for reliable and repeatable interpretation of musculoskeletal imaging studies. Furthermore, the diagnostic performance of these deep learning methods must be evaluated in prospective studies using a large number of image datasets acquired at different institutions using different imaging parameters and different imaging hardware before they can be implemented in clinical practice.

Acknowledgements

No acknowledgement found.