ANALISA KERAPATAN TRABECULAR BONE BERBASIS GRAPH BERBOBOT PADA CITRA PANORAMA GIGI UNTUK IDENTIFIKASI OSTEOPOROSIS

Abstract

Osteoporosis is bone disease indocated by low bone mass density and micro architectures disorder which lead to bone fragility or fractures. Graph may be useful to describe density of trabeculae bone due to morphological change on mandibular bone in dental panoramic radiographs. The density of trabecular bone can be discribed by generating graph. Trabecular image firstly was transformed to binary image. A white pixel on the binary image presented as part of trabeculae, which assumed as an isolated node on the graph. Graph generation by Erdos and Royi method was used to build connections between an isolated node and others. This paper introduced the use of weight on each node based on probabilities average of its neigbourhoods.

Graph’s properties which used to measure the density were degree and cluster coefficient. Both of properties are used to build feature space. Feature space indicated distribution of node on dense or sparse area. Early indication of osteoporosis could be assumed that ratio of nodes on dense area were greater than that on sparse area. We achieved accuration of 54%, sensitivity of 60%, and spesificity of 49%.

Keywords: osteoporosis, trabeculae, random graph, graph berbobot