Automated knee cartilage segmentation for heterogeneous clinical MRI using generative adversarial networks with transfer learning.

Background: This study aimed to build a deep learning model to automatically segment heterogeneous clinical MRI scans by optimizing a pre-trained model built from a homogeneous research dataset with transfer learning. Methods: Conditional generative adversarial networks pretrained on the Osteoarthritis Initiative MR images was transferred to 30 sets of heterogenous MR images collected from clinical routines. Two trained radiologists manually segmented the 30 sets of clinical MR images for model training, validation and test. The model performance was compared to models trained from scratch with different datasets, as well as two radiologists. A 5-fold cross validation was performed. Results: The transfer learning model obtained an overall averaged Dice coefficient of 0.819, an averaged 95 percentile Hausdorff distance of 1.463 mm, and an averaged average symmetric surface distance of 0.350 mm on the 5 random holdout test sets. A 5-fold cross validation had a mean Dice coefficient of 0.801, mean 95 percentile Hausdorff distance of 1.746 mm, and mean average symmetric surface distance of 0.364 mm. It outperformed other models and performed similarly as the radiologists. Conclusions: A transfer learning model was able to automatically segment knee cartilage, with performance comparable to human, using heterogeneous clinical MR images with a small training data size. In addition, the model proved robust when tested through cross validation and on images from a different vendor. We found it feasible to perform fully automated cartilage segmentation of clinical knee MR images, which would facilitate the clinical application of quantitative MRI techniques and other prediction models for improved patient treatment planning.

Analysis of Different Levels of Structured Reporting in Knee Magnetic Resonance Imaging.

RATIONALE AND OBJECTIVES: The aim of this study is to investigate the most appropriate knee MRI report template that not only provides structure and consistency, but also allows enough narrative freedom for the logical organization of findings and improved communication with the orthopedic referral base. MATERIALS AND METHODS: Three fictitious knee MRI reports were created using templates with different levels of structuring: unstructured free text (FT), structured with headers (SH), and highly structured and itemized (SI). These were then distributed to clinicians in the orthopedics department at all levels of training along with a survey with numerical scoring questions on report readability, usefulness, and quality. Statistical analysis was used to evaluate the data. RESULTS: Fifty-three surveys were completed with responses from residents, attendings, and physician assistants. The structured format with headers had statistically significant (p value <0.001) higher mean rank score in readability, usefulness, and quality parameters compared to the unstructured FT and highly SI report templates. Most clinicians (83%) found the structured format with headers to be the most coherent report. Conversely, 53% found the unstructured FT and 43% found the highly SI templates to be the most disjointed. CONCLUSION: Based on responses to surveys of knee MRI report templates, our results show that our orthopedic clinicians prefer some level of structure in the reports but not the rigorous itemization of anatomic tissues. A "middle ground" reporting structure which includes headers for different anatomic compartments and allows for grouping of relevant pathology, is shown to be the preferred format.