To explore the application value of APP-based self-learning and teaching mode in small joint ultrasound training.

Thirteen sonographers were included in this study. First, they were allowed to learn "the joint ultrasound scoring system" independently through the electronic course set in the APP. Following the course, they were admitted to enter "the APP ultrasonographic image assessment platform" to perform ultrasound scoring for 140 small joints with inflammatory rheumatic diseases in the form of multiple choices. The ultrasound score included the second and third metacarpophalangeal joints (MCP2/3), wrist joints, and the second and third proximal interphalangeal joints (PIP2/3), which were evaluated for synovitis, tenosynovitis/paratenonitis, and erosions from the dorsal side by gray-scale ultrasound (GSUS) and power Doppler ultrasound (PDUS). The overall agreement rate (%) and intraclass correlation coefficient (ICC) values were calculated to analyze the ultrasound training effect.

A moderate to good interobserver agreement was observed for the detection of synovitis, tenosynovitis, and erosions (ICC=0.49-0.97, overall agreement 76.4%~90.4%). In detail, the ICC value for GSUS in evaluating synovitis was only 0.65 (overall agreement 45.2%), and for PDUS it was 0.88 (overall agreement 75.7%). The best training results were found for the detection of synovitis and erosions by PDUS in the MCP2 joint (80.8%/0.77; 88.5%/0.95). Good ICC values were observed for GSUS in the detection of synovitis in the MCP2/3 joints; however, the overall agreement rates were lower (0.63, 0.75; 44.4%, 46.2%). Poor agreement values were observed for GSUS in detecting erosions in both the wrist region and synovitis in PIP2/3 joints (27.4%-53.8%; 0.14-0.37). Good overall agreement rates were observed for GSUS in the detection of synovitis in the wrist, and erosions in PIP2/3 and MCP3 joints, while the ICC values were lower (74.40%-81.2%; 0.24-0.38). The overall ultrasound training effect was similar between MCP2 and MCP3 joints (71.8%/0.83 vs 64.7%/0.83), as well as between PIP2 and PIP3 joints (58.1%/0.60 vs 59.3%/0.45).

APP-based self-learning and testing mode has the potential to be an effective tool in joint ultrasound training. More training and standardization of image acquisition, and adequate interpretation for multiple joints with poor agreement values will help to further optimize the teaching effect and improve the clinical diagnostic ability.