自己組織化マップを応用した顎顔面の個成長予測

概要

論文の詳細を見る
Considerations of individual growth characteristics of the jaws are of great importance to make treatment plans for individual orthodontic patients during their growth. The purpose of the present study was to apply a self-organizing map (SOM) to an individual growth prediction of the face and to evaluate the reliability of this prediction. The longitudinal lateral cephalograms of 44 girls taken three years before and one year after the age of maximum puberal growth, denoted stage A and stage B respectively, were examined. All cephalometrically measured values in an individual subject were converted into Z scores in relation to the mean and standard deviation (SD) of the respective measurements in the corresponding stage, to obtain input vectors for the SOM. By repeatedly inputting the stage B vectors obtained from the 44 subjects into the initialized SOM with 10×10 processing units, self-organization of the units gradually proceeded, and a final map was obtained for a growth prediction after one million learning events. The 100 units on this prediction map corresponded to the virtual craniofacial patterns of 100 individuals. Based on the morphological similarities of the face throughout the growth period, the facial pattern to be predicted was determined among the units by inputting the vector obtained from the stage A data into the prediction map for each individual. The average difference between the predicted and actual facial forms was 0.35 SD per measurement. According to Moyer's criteria, clinically satisfactory prediction results were obtained for 75% of the subjects, whereas the remaining 25% had less predictable forms. These results suggest that the SOM has the potential to predict craniofacial growth for individuals.
福岡歯科大学学会の論文
2002-12-20

自己組織化マップを応用した顎顔面の個成長予測

スポンサーリンク

概要

著者

関連論文

スポンサーリンク