Title: A performance-based features selection for automatic identification of bone X-ray images

Authors: Mrityunjaya V. Latte; V. Kumar Swamy; Basavaraj S. Anami

Addresses: JSS Academy of Technical Education, Uttarahalli Kengeri Road, Bengaluru 56006, Karnataka, India ' K.L.E. Institute of Technology, Gokul Road, Hubli 580030, Karnataka, India ' K.L.E. Institute of Technology, Gokul Road, Hubli 580030, Karnataka, India

Abstract: This paper presents a performance-based features selection for automatic identification of bone X-ray images using Artificial Neural Network (ANN). Orthopaedicians use imaging technology such as X-ray, computer tomography, magnetic resonance imaging and the like. In this work, we have considered images of different bone types in the human skeleton such as skull, ribs, tibia, femur, foot, hand, ulna, wrist and the humerous using X-ray images. The shape and texture features are selected based upon the identification accuracies. ANN is used for identification. Identification accuracy of different bones using shape, textures, combined shape and texture features are analysed. It is observed that the accuracies for shape, texture and their combination are 52.87%, 32.33% and 95% respectively. The work is useful for orthopaedics practitioners.

Keywords: digital radiography; orthopaedics; image identification; artificial neural networks; ANNs; feature selection; automatic identification; autoID; bone x-rays; x-ray images; shape features; texture features.

DOI: 10.1504/IJBET.2013.057264

International Journal of Biomedical Engineering and Technology, 2013 Vol.12 No.3, pp.252 - 276

Received: 09 Apr 2013
Accepted: 01 Aug 2013
Published online: 27 Sep 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article