Computerized System for Screening Aged Women with Low Bone Mass Using Digital X-ray of Calcaneum
Amrita B Pal1, Priyam Singh2
1Amrita B Pal, Deputy Manager, Operations Sunshine Hospital Bhubaneswar (Odisha), India.
2Priyam Singh, Assistant Professor, Department of ECE, Vignan Foundation for Science, Technology Research, Guntur (Andhra Pradesh), India.
Manuscript received on 14 February 2019 | Revised Manuscript received on 05 March 2019 | Manuscript Published on 08 June 2019 | PP: 358-362 | Volume-7 Issue-5S4, February 2019 | Retrieval Number: E10750275S419/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Low bone mass (LBM) is a universal health problem in which the bone becomes fragileand more frequent in women than the men. The objective was to evaluate the adequacy of the plain digital X-ray image of calcaneum for the low bone mass evaluation by implementing neural network with a feasible accuracy when compared to X-ray with dual energy absorptiometry. Here for the study purpose, total women studied (n=52, aged 30 years and above) were classified as follows: Group-I: Normal (n=26), Group-II: Women with LBM (n=26). In each subject, a X-ray was taken for right calcaneum lateral viewn. Also, we measured bone mineral density for right proximal femur by using DXA. X-ray image was processed in MATLAB tool. A semi-automatic technique is been employed for selecting the area with calcaneum, and its trabeculae features were extracted using Canny detection technique, shape features, texture analysis, and gray level co- occurrence matrix. The feature selection was done, based on high value (≥0.6) of measure of sample adequacy (MSA) of features using principal component analysis (PCA). The classification using selected features was done with the help of an artificial neural network (ANN). In women with LBM (Group-II), the mean values of number of white pixels, solidity and contrast of calcaneum were lesser significantly, when compared to the corresponding values measured in normal women (Group-I). A semi-automatic computer aided diagnosis (CAD) tool was developed to evaluate LBM from digital X-ray of calcaneum using ANN. The accuracy of the tool was found to be 94.2%, when compared to DXA. Hence, calcaneum X-ray can be used as a inexpensive technique for evaluation of LBM.
Keywords: Low Bone Mass, Bone Mineral Density, Dual-Energy X-ray Absorptiometry, Artificial Neural Network.
Scope of the Article: Digital Signal Processing Theory