Study On The CT Image Characteristics About The Metastatic Lymph Node In N Staging Of Lung Cancer

This paper put forward a new method, in which the texture characteristics were characteristic vectors of lymph nodes in CT images, to diagnose metastatic lymph nodes for solving difficult problems between cancerous tissue pathology characteristic and CT image texture in N staging of lung cancer.Firstly, we studied the optical properties of biological tissue based on the understanding CT imagings. The intensity distribution of X-ray was simulated in normal tissue mixed with cancerous tissue. Through attenuation extent of normal and cancerous tissue in X-ray, we illustrated that CT image textures can be used as medical diagnostic criteria.Secondly, graylevel co-occurrence matrix and multiresolution graylevel histogram methods were adopted to extract texture feature of lymph node in chest CT images.Finally, diagnose metastatic lymph nodes in CT images can boil down to image classification issue. Based on SMO method SVM classifier was constructed diagnosis model with LIBSVM software because SVM had good generalization ability in image classification.The texture features extracted by graylevel co-occurrence matrix, simple graylevel histogram and multiresolution graylevel histogram method were input characteristic vectors of lymph node characteristics in CT images.Based on the training and testing of SVM classifiers, diagnostic accuracy can reach 70% of SVM classifiers with multi-resolution graylevel histogram method, and obviously were higher than using other methods.Texture feature of lymph nodes in chest CT images with multiresolution graylevel histogram method can be a new diagnosis basis of metastatic lymph nodes in N staging of lung cancer.