Spectrum identification of hepatitis tongue diagnosis based on principal component analysis and BP neural network

To achieve non-invasive detection on hepatitis patients, a model was established for identification of this disease based on normalized data of reflective spectrum from both healthy adults and viral hepatitis suffers, which was processed by the method of principal component analysis(PCA) and BP neural network. Two subsets of spectrum data containing 36 samples each were acquired from tip position of tongues of healthy adults and viral hepatitis patients respectively. Since being normalized, PCA was conducted to acquire principal components. Eight principal components(PCs) were selected based on accumulative reliabilities, and these selected PCs would be taken as the inputs of artificial neural network. Then 26 samples from each subset were selected randomly to make a modeling dataset utilized by a ANN, and the rest for testing samples. The result shows a 100% recognition correction under the condition that a predictive error threshold of ±0.2 is set. The experimental result shows that the PCA-BP method can achieve good classification and recognition on features of healthy people and viral hepatitis patients, which greatly promotes the objectiveness of traditional Chinese medicine tongue diagnosis.