Volume : VII, Issue : XII, December - 2018
A Support Vector Machine Based Technique for Early Stage Osteoporosis Classification from Pelvic Bone X-Ray Images
Raghavendra Chinchansoor, Prof. And Chairperson
Abstract :
Osteoporosis is one of the very common pathophysiological traits which is getting observed in the modern urban population. Although over the years, several sophisticated diagnostic techniques are being developed to effectively detect Osteoporosis, the adaptation is not wide due to time, money and skill complexities with these methods. X–ray is one of the most widely available medical imaging methods but it has been traditionally used in detecting fractures and other severity in the bone structure. Osteoporosis deals with weakening of the bones which are not essentially always get detected through X–ray at the early stage. In this paper we present an efficient technique for early stage Osteoporosis detection with an accuracy of 84% that leverages the texture and structure properties of the pelvic section X–ray image, a prior knowledge and binary SVM classifier to classify a given pelvic section X–ray of a subject to asses if that subject has Osteoporosis or not. Our method relies on extracting high dimensional feature vectors using fractal dimension from the bone X–ray image and training a SVM classifier with linear kernel to classify the image data into normal or Osteoporosis
Keywords :
Medical imaging Neural Network SVM Femur X–Ray Osteoporosis Fracture
Article:
Download PDF
DOI : https://www.doi.org/10.36106/paripex
Cite This Article:
A Support Vector Machine Based Technique for Early Stage Osteoporosis Classification from Pelvic Bone X-Ray Images , Raghavendra Chinchansoor, Prof. And Chairperson , PARIPEX-INDIAN JOURNAL OF RESEARCH : Volume-7 | Issue-12 | December-2018
Number of Downloads : 140
References :
A Support Vector Machine Based Technique for Early Stage Osteoporosis Classification from Pelvic Bone X-Ray Images , Raghavendra Chinchansoor, Prof. And Chairperson , PARIPEX-INDIAN JOURNAL OF RESEARCH : Volume-7 | Issue-12 | December-2018