medwireNews: Osteoscan, an artificial intelligence program, determines the radiographic stage of knee osteoarthritis with 95% accuracy, indicate findings from the EULAR 2020 E-Congress.
The technology determines the stage by analyzing radiographic images of the knee joints uploaded by a rheumatologist as a photo or via DICOM and their responses to questions on clinical manifestations in “difficult cases,” presenter Olga Georginova (Moscow State University, Russian Federation) told delegates.
For the current study, the neural network was trained on an initial set of 1300 X-rays of both knee joints – 350 were photos and 950 in the DICOM format. From these, a quality control sample of 130 images was selected and the technology’s ability to determine osteoarthritis stage was tested against that of an independent group of certified radiologists (considered the gold standard).
In 124 out of 130 cases, the radiologist was in agreement with the system in determining osteoarthritis stage, which represents an accuracy of 95.4%, reported Georginova.
medwireNews is an independent medical news service provided by Springer Healthcare. © 2020 Springer Healthcare part of the Springer Nature Group