Tumor cell image recognition based on PCA and two-level SOFM

Abstract

In this paper, a method based on PCA and two-level SOFM neural network is proposed for tumor recognition. The method combines PCA with a two-level SOFM neural network in which PCA is used to reduce the dimensionality of the input tumor image sample and the two-level SOFM neural network is used to extract characters and classifying. This method compromises linear dimensionality reduction, character extraction and classification. The training learning of the tumor image samples in the clinical pathological diagnosis can get the parameter of the two-level SOFM neural network. The experiment shows that the proposed method has better classifying accuracy and the classifying time is letter than the other methods such as PCA, LLE, PCA+LDA, SVM and two-level SOFM.