An automatic model-based system for joint space measurements on hand radiographs: initial experience

Area of Science:

• Radiology
• Medical Imaging
• Rheumatology

Background:

• Accurate measurement of joint space width is crucial for diagnosing and monitoring arthritic conditions.
• Manual radiographic analysis can be time-consuming and subject to inter-observer variability.

Purpose of the Study:

• To evaluate the performance of a novel web-based service for automated joint space width measurement in metacarpophalangeal joints.
• To assess the reproducibility and accuracy of the automated system compared to a reference standard.

Main Methods:

• Prospective testing of a web-based service for automated radiographic analysis.
• Measurement of 160 metacarpophalangeal joint spaces in 20 patients with suspected rheumatoid arthritis or osteoarthritis.
• Comparison of automated measurements with interactive segmentation as the reference standard.

Main Results:

• The automated system achieved a 93% technical success rate.
• Smallest detectable differences ranged from 0.08 to 0.31 mm, with a coefficient of variation of 2%-7%.
• Mean error compared to the reference standard was 0.19-0.40 mm.

Conclusions:

• The developed web-based service provides reproducible joint space measurements in metacarpophalangeal joints.
• The automated system demonstrates potential for efficient and accurate analysis of radiographs in patients with moderate erosive and osteophytic disease.
• This technology may aid in the diagnosis and management of rheumatoid arthritis and osteoarthritis.