Background Though musculoskeletal ultrasonography (MSUS) is largely recognized as a useful imaging modality to enhance disease assessment and management in patients with rheumatoid arthritis (RA), it remains a challenge that MSUS are poor objectivity and reliability in evaluating RA lesions.
Objectives To improve objectivity of MSUS, we developed the two programs; automatic acquisition of the most appropriate Power Doppler (PD) gain (Quick Scan) and automatic extraction of appropriate image. In this study we assessed the efficacy and validity of these programs.
Methods Joints of RA patients were evaluated with Aplio XG possessing the two new programs by 2 experts in MSUS blindly each other. First, to assess validity of Quick Scan, the variation in PD gain values obtained by using Quick Scan between two examiners was compared to those obtained manually. Next, one examiner conventionally adjusted PD gain and captured appropriate still images which showed maximum PD signals indicating synovitis, whereas the other examined the same joints by using Quick Scan and storing movies of a whole extent of the joints. According to an algorithm that chooses the image of maximum PD pixel ratio after excluding occasions of increased PD signals by moving noise or normal blood flow, several images were automatically extracted from movies. PD scores of still and moving images as well as those of the automatically-extracted images were evaluated later by another expert. PD signals were graded from 0 to 3 with semi-quantitative method in each joint.
Results Seventeen cases, 62 joints, and 117 lesions were assessed. After pushing key of Quick Scan, PD gain was adjusted automatically within a second. The evaluation of PD gain adjustment was done in 30 joints and the gain values obtained by Quick Scan tended to be more similar between the examiners (Δ1.33±0.92) than those adjusted manually (Δ2.60±4.47) (p=0.067). PD scores evaluated from moving images and those of the automatically-extracted images from the movies were well matched (κ coefficient 0.85) and 75% of the automatically-extracted images coincided with the sites of movies which the expert chose for evaluating PD score. Concordant rate between PD scores of conventionally-captured images and those of the automatically-extracted images was also good (κ coefficient 0.74). These results showed that the image-extracted algorithm as well as usage of two programs to evaluate synovitis was considered reasonable and proper. Moreover, the scores of automatically-extracted images were higher than those of still images in 17 joints, which suggested that the appropriate PD gain obtained by Quick Scan and automatic extraction program without subjective eyes may contribute to prevent the examiners from missing PD signals.
Conclusions These new programs enable MSUS operators to evaluate arthritis objectively, efficiently, and accurately.
Disclosure of Interest None Declared