The next generation of machine learning can be harnessed to interpret knee MRI scans quicker and as accurately as clinical experts.
A Stanford University team of computer scientists, radiologists and surgeons have developed a deep learning model for detecting abnormalities such as ACL and meniscal tears.
MRNet is a convolutional neural network that can take a series of MRI images and ‘learn’ to identify abnormalities.
Their study used 1,370 mostly abnormal knee MRI scans and compared the model’s interpretation to that of the clinical experts. The study also validated MRNet on a second cohort of 917 scans from Croatia.
The model performed as well as experienced clinicians with area under the receiver operating characteristic curves (AUCs) of 0.937, 0.965 and 0.847 for abnormality detection, ACL tear detection and meniscal tear detection respectively.
“Notably, the model achieved high specificity in detecting ACL tears on the internal validation set, which suggests that such a model, if used in the clinical workflow, may have the potential to effectively rule out ACL tears,” the study authors said.
“We also found that providing the deep learning model predictions to human clinical experts as a diagnostic aid resulted in significantly higher specificities in identifying ACL tears.”
“Finally, in contrast to the human experts, who required more than three hours on average to completely review 120 exams, the deep learning model provided all classifications in under 2 minutes.”
“Our results suggest that deep learning can be successfully applied to advanced musculoskeletal MRI to generate rapid automated pathology classifications and that the output of the model may improve clinical interpretations.”
They added it was possible that with enough ‘normal’ training data to compare to, a deep learning model could be trained to detect even uncommon abnormalities.
“Future studies are needed to improve the performance and generalisability of deep learning models for MRI and to determine the effect of model assistance in the clinical setting.”
The findings are published in PLOS Medicine