Premium
Probabilistic analysis of peri‐implant strain predictions as influenced by uncertainties in bone properties and occlusal forces
Author(s) -
Petrie Cynthia S.,
Williams John L.
Publication year - 2007
Publication title -
clinical oral implants research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.407
H-Index - 161
eISSN - 1600-0501
pISSN - 0905-7161
DOI - 10.1111/j.1600-0501.2007.01384.x
Subject(s) - cancellous bone , premolar , cortical bone , implant , materials science , probabilistic logic , dentistry , probabilistic analysis of algorithms , modulus , mandible (arthropod mouthpart) , finite element method , orthodontics , mathematics , composite material , structural engineering , medicine , statistics , molar , surgery , engineering , anatomy , botany , biology , genus
Objectives: The purpose of this study was to examine the influence of variability in bone properties and loading on peri‐implant crestal and cancellous bone strains using a probabilistic approach, in combination with finite element (FE) analysis. Material and methods: Oblique occlusal loading was applied to a single endosseous implant embedded in a two‐dimensional (2‐D) FE model of a premolar section of a mandible. Perfect bonding was assumed at all interfaces. Five independent parameters (cortical bone thickness (T), cortical (ECORT) and cancellous (ECANC) bone Young's moduli, and vertical (FVERT) and horizontal (FHOR) occlusal forces) were assigned statistical distributions based on data in the literature. Two cancellous bone distribution models were examined, one with a lower mean and range (LM) and the second with a higher mean and range (HM) of cancellous bone Young's modulus values. Sets of randomly chosen values for the five parameters were selected from the distributions and FE analyses were performed for all randomly selected sets. Results: In the LM model, 50% of the cases experienced hyper‐physiologic peri‐implant crestal strains in the region of commonly reported saucerization, compared with about 25% of the cases in the HM model. Relative probabilistic sensitivities (%) of bone strains to the independent input parameters (T, ECORT, ECANC, FVERT and FHOR) were as follows: 29, 11, 30, 13, and 17 for the LM model, and 17, 11, 35, 21, and 15 for the HM model, respectively. Conclusion: Probabilistic analyses of FE models suggest that up to twice as many cases in the LM distribution may be at risk of saucerization as compared with the HM distribution model. Although based on hypothetical distribution values and the limitations inherent to a 2‐D analysis, this probabilistic study demonstrated that FE models are very sensitive to the often arbitrarily assigned values for cancellous bone Young's modulus, and also to values used for cortical bone thickness, when the cancellous bone modulus is low.