Development and validation of an artificial intelligence system based on deep algorithms for diagnosing metacarpophalangeal joint lesions in gouty arthritis using high-frequency musculoskeletal ultrasound
Objective To develop an artificial intelligence(AI)system based on deep algorithms for di-agnosing metacarpophalangeal(MCP)joint lesions in gouty arthritis using high-frequency musculoskeletal ul-trasound and to validate its diagnostic performance.Methods High-frequency musculoskeletal ultrasound im-ages of the hands of 410 patients with gouty arthritis who underwent ultrasonography in our hospital from February 2017 to December 2023 were retrospectively collected and divided into a training set(n=287)and a validation set(n=123)in a 7:3 ratio.An AI system for diagnosing MCP joint lesions in gouty arthritis was developed using deep algorithms and trained with the training set.With positive polarized light examination of MCP joint synovial fluid as the diagnostic criterion,the diagnostic performance of junior and senior ultrasound physicians and the AI system for MCP joint lesions in gouty arthritis in the validation set was compared.Re-sults Among the 410 patients with gouty arthritis,343 had MCP joint lesions,with an incidence rate of 83.66％ .The sensitivity,specificity,accuracy,and area under the receiver operating characteristic curve(AUC)of the AI system for diagnosing MCP joint lesions in the validation set were 85.29％,85.71％,85.37％,and 0.855,respectively.The sensitivity and accuracy of the senior ultrasound physicians in diagno-sing MCP joint lesions in the validation set were higher than those of the junior ultrasound physicians,with statistically significant differences(P＜0.05).The sensitivity and accuracy of the AI system in diagnosing MCP joint lesions in the validation set were higher than those of the junior ultrasound physicians but lower than those of the senior ultrasound physicians,with statistically significant differences(P＜0.05).However,there was no statistically significant difference in sensitivity between the AI system and the senior ultrasound physicians(P＞0.05).Conclusion The AI system developed using deep algorithms has good diagnostic per-formance for MCP joint lesions in high-frequency musculoskeletal ultrasound images of patients with gouty arthritis and provides higher diagnostic value compared to junior ultrasound physicians.
Deep algorithmsHigh-frequency musculoskeletal ultrasoundGouty arthritisMetacarpophalangeal joint lesionsArtificial intelligence
万方数据
维普
