Molecular Structure Images Classification by Fuzzy-DeepLearning for Early Diagnosis of Gastric Cancer

Early diagnosis is very important in reducing the death rate of cancer patients. Therefore, in this paper, anintelligent method has been developed for early detection of gastric cancer. The approach of this researchis to use the Rough-Score feature extraction method, which is followed by classification with the DeepPyramid-Fuzzy Neural Network. The results of the proposed method used some other dimensionreduction methods with Rough-Score in feature extraction which consider Local Binary Patterns (LBP),Histogram of Oriented Gradients (HOG), Sammon Mapping, ISOMAP, classical Multi-DimensionalScaling (MDS), Local Linear Embedding (LLE), Linear Discriminant Analysis (LDA), t -DistributedStochastic Neighbor Embedding (t-SNE). New medical systems have been developed to measure theeffects of these dimension reduction methods by obtaining features exaction with Rough-Score indifferent dimensions. The final results represented the application of the Rough-Score feature extractionmethod with the MDS dimension reduction method has a higher accuracy in dimensions reduction of۵x۱۸۰ and leads to a better classification of molecular structure for the gastric cancer diagnosis with deeppyramid-fuzzy neural network classification.

کلیدواژه ها:

Gastric Cancer ، Molecular Structure ، Rough-Score ، Deep Pyramid-Fuzzy NeuralNetwork