Osteoporosis Classification Using Multifractal Features and Neighbourhood Determination of Extrema for BMED

Abstract

Reduced bone mineral density and abnormalities in the microarchitecture of the bone are signs of osteoporosis, which makes the bones more susceptible to breaking and increases the risk of multiple fractures. One major challenge in the analysis and classification of medical pictures is the use of radiography images to differentiate between osteoporotic persons and those who are healthy. To characterize the trabecular bone microarchitecture of patients with osteoporosis, this work presents a novel method based on the analysis of the multifractal texture of the preprocessed proximity pictures of the BMED decomposition Neighbourhood Determination of Extrema. With the suggested technique, radiographs can describe the trabecular bone’s regularity and roughness. The multifractal spectrum serves this function by highlighting alterations in bone microarchitecture brought on by osteoporosis and identifying novel textural features. An SVM model was used to conduct tests to classify two groups. A 99.11% score was recorded, suggesting that this approach could be useful as an extra tool for osteoporosis diagnosis.