Fruit fly optimization algorithm based modified least squares twin SVM for grading of metacarpophalangeal rheumatoid arthritis

Abstract

In the past few decades, ultrasound (US) has witnessed a drastic progress in the inflammatory joint diseases field. RA (Rheumatoid arthritis) is possibly the pathology, which has gained the most advantages due to this progress, both with respect to timely diagnosis and monitoring of disease course. Training a professional sonographer is costly and time- constrained whereas some studies pay attention towards automated RA grading techniques and itis possible using machine learning (DL) technique, to yield a highly objective, automated and rapid means of RA diagnosis in clinical environments. Hence, effective MLTs (machine learning techniques) that are less time consuming while achieving better performances in quick recognition of RA diseases should be proposed by combining soft computing methods. With this motivation, this work proposes FFO-MLSTSVM (Fruit Fly Optimization Algorithm Based Modified Least Squares Twin Support Vector Machines) method for Grading of MRA (Metacarpophalangeal RA). This technical work makes use of the parameters and texture characteristics pertaining to ROIs (regions of interest).