Abstract #0201

Automated Knee Cartilage Segmentation with Very Limited Training Data: Combining Convolutional Neural Networks with Transfer Learning

Alexander R Toews^1,2, Zhongnan Fan³, Marianne S Black^2,4, Jin Hyung Lee^1,3,5,6,7, Garry E Gold^2,5,8, Brian A Hargreaves^1,2,5, and Akshay S Chaudhari^2,5

¹Electrical Engineering, Stanford University, Stanford, CA, United States, ²Radiology, Stanford University, Stanford, CA, United States, ³LVIS Corporation, Palo Alto, CA, United States, ⁴Mechanical Engineering, Stanford University, Stanford, CA, United States, ⁵Bioengineering, Stanford University, Stanford, CA, United States, ⁶Neurology & Neurological Sciences, Stanford University, Stanford, CA, United States, ⁷Neurosurgery, Stanford University, Stanford, CA, United States, ⁸Orthopaedic Surgery, Stanford University, Stanford, CA, United States

Magnetic resonance imaging is commonly used to study osteoarthritis. In most cases, manual cartilage segmentation is required. Recent advances in deep-learning methods have shown promise for automating cartilage segmentation, but they rely on the availability of large training datasets that rarely represent the exact nature or extent of data practically available in routine research studies. The goal of this study was to automate cartilage segmentation in studies with very few training datasets available by creating baseline segmentation knowledge from larger training datasets, followed by creating transfer learning models to adapt this knowledge to the limited datasets utilized in typical study.

This abstract and the presentation materials are available to members only; a login is required.

Join Here