Diagnose flat foot from foot print image based on neural network

Abstract

Normally, there have been many methods to diagnosis of flat foot. Each method is different to use indicators e.g. Staheli arch index, Clark's angle and Chippaux-Smirak index. However, the results from such indicators are still varied in each method. Therefore, this paper proposes a classification of the flat foot by combining of multiple indicators with neural network process. It can improve an accuracy of classification more than the use of only one indicator. There are 132 images of footprints (left and right foot) consisting of normal foot or flat foot. The experimental results using a combination of indicators show that an accuracy of the result is up to 93% more than the single index i.e. Staheli arch index 43%, Clark's angle 68%, Chippaux-Smirak index 80%. It can make more precisely diagnose of flat foot. © 2013 IEEE.