Histogram Based Synovitis Scoring System using Ultrasound Images of Rheumatoid Arthritis
LC10-LC14
Correspondence
Dr. V Vijayabaskar,
Professor and Head, Department of Telecommunication Engineering, Sathya Bama University, Chennai-600119, Tamil Nadu, India.
E-mail: hodbiomedical@velsuniv.ac.in
Introduction: The gray scale Ultrasound (US) imaging aids in assessing activity of Rheumatoid arthritis condition through classification of synovitis and is able to estimate the disease progression. The gray shades in the synovial region help in identifying the disease condition.
Aim: To grade rheumatoid arthritis based on the severity of the disease condition using histogram analysis.
Materials and Methods: A total of 276 ultrasound gray scale images of the arthritis affected finger joints from MEDUSA database were evaluated. The image was preprocessed using filtering to remove speckle noise and the quality was enhanced using CLAHE technique. The region of interest (synovial region) was analysed using histogram and the respective histogram features were obtained. Based on the histogram analysis the images were graded into different grades. The histogram analysis was applied to reference image (annotation done by medical expert). A comparative study between the reference images and test images were performed.
Results: Based on the histogram analysis the images were graded into Grade 0- no disease, Grade 1- less severity, Grade 2- moderate, Grade 3- more severity. The performance of the histogram method on reference and test image was compared. Using the histogram analysis, 93% sensitivity, specificity of 94% and an accuracy of 93.5% were achieved.
Conclusion: Ultrasound image of synovial thickness is an important tool in identifying early stages of rheumatoid arthritis. The histogram based scoring approach gives gray level variations in US image. Based on this mean, median and standard deviations are calculated. Average value of all these parameters was very high for Grade 0, and very low for Grade 3. The outcome of suggested system demonstrates effective and satisfactory performance for different set of synovitis images of RA patients; which would be helpful for early diagnosis of the disease.