The Research And Design On Diagnosis System Of Cervical Smear Based On Neural Network

At present, digital images processing, pattern recognition and artificial intelligence are widely applied in the fields of biology and medicine. Some achievements have been gained by these techniques. But there is few reports on the pathologic diagnosis system which can be used to recognize cells images in the mean of neural network. In the paper, based on the detailed investigation of cervical cells images recognition and feature extraction, results of utilizing neural network to diagnose and classify cells images are satisfiable and valuable in clinical application.In the paper, the original cervical cells images are preprocessed and samples are gained. Features parameters of morphology and chroma are extracted from images of cells samples. The images of cells samples are recognized and classified by MLPNN (Multilayer Perceptron Neural Network) and RBFNN (Radial Basis Function Neural Network). Several arithmetics of MLPNN and RBFNN are discussed, and cross entropy arithmetic are suggested. Among the recognized results, the recognition rate and classification of RBFNN and MLPNN with BP arithmetic based on cross entropy are the best one.Cervical smear recognition system bases on neural network is designed and developed. The system can be used to merge, classify and standardize files of the cell feature parameters automatically. The standardized feature parameters file can be recognized and classified by different neural network files, and the results are saved in a output file. The cell image according with its diagnosed record can be previewed.Combined computer technology with practical experiences of pathology experts and based on processing techniques of medicine images, neural network can be used to recognize cells images and shows its significant value in clinical diagnosis and medical research at present and in the future.