Inflammatory Bone Region Segmentation using USG Rheumatoid Arthritic Images

Accurate Inflammatory bone region localization gives significant clues about the presence of Rheumatoid Arthritis (RA). Rheumatoid Arthritis mostly affects the smaller joints like finger and wrist joints in the early phase of the disease. Disease detection in the initial phase, is the exclusive way towards better treatment outcomes and, in the long run, reduces the degenerative effects of the disease. Advancement of the digital imaging modalities like Ultrasonography is very significant for small joint analysis like different types of hand joint those who are commonly affected by RA in early phase. The segmentation of musculoskeletal ultrasonography images is the primary important step in the early diagnosis of illnesses. In this paper, we investigated the efficiency of quality enhancement of musculoskeletal ultrasound images for segmentation of inflammatory bone regions towards diagnosis of arthritis. To meet up the specific necessities of the overall aim of this paper, firstly, various state-of-the-art quality enhancement techniques (i.e., noise reduction techniques) are implemented for quality improvement. Secondly, the performance of the state-of-the-art segmentation methods to extract the inflammatory bone line/ region from ultrasound images are investigated in terms of before and after quality enhancement. The experiments are conducted on publicly available ultrasound image dataset of arthritis patients available at: https://www.ultrasoundcases.info/. Experimental results reveal that geometric mean filtering technique (i.e., best performed enhancement technique) improves the performance of the segmentation methods for bone line segmentation.