| Whether the patients with colon cancer have liver metastasis has a significant impact on their survival rate and prognosis.However,at present,the early diagnosis rate of patients is low,the survival period is short,and the diagnosis is not optimistic.Therefore,taking effective measures to evaluate the risk of liver metastasis and control the development of tumor in advance will help clinicians to carry out effective personalized prognosis.The purpose of this paper is to propose the median method based on enhanced CT images and M-K means clustering analysis based on enhanced CT images to identify the risk of liver metastasis in patients with colon cancer,and to compare the performance of the two methods in the training set and the test set by using the confusion matrix,so as to explore the independent prognostic factors of the risk of liver metastasis in colon cancer cells,and to select the most potential risk stratification scheme.59 cases of colon cancer admitted to hospital from 2018 to 2019 were examined by enhanced CT,and gray-scale arrays of lesions in arterial phase and portal phase of CT sequence are extracted respectively to obtain the two-dimensional CT value characteristics of each pixel in the tumor.Then,by using the median method based on CT enhancement images and the M-K means clustering method based on CT enhancement images after optimization,the tumor focus is divided into fourmicro-environment sub-regions with different aggressiveness by combining different periods and CT values.We select the highly aggressive micro-environment "Habitat" sub-region with high CT value in arterial phase and low CT value in portal phase,and predict the risk of liver metastasis according to the number of pixels gathered in this sub-region.The diagnostic accuracy,sensitivity,specificity,ROC curve and AUC value of the two methods in the test set are calculated to determine whether the clustering algorithm can effectively evaluate the highly aggressive micro-environment sub-region in the tumor,and then become a prognostic factor for the risk of liver metastasis.The results show that the M-K means clustering method based on CT enhanced images has higher accuracy,diagnostic sensitivity and specificity parameters,and lower misdiagnosis rate and missed diagnosis rate compared with the median method based on CT enhanced images,and in the ROC curve,AUC values are 0.848 and0.709 respectively,the former has higher diagnostic accuracy.Therefore,M-K means cluster method based on CT enhanced image can accurately evaluate the risk of liver metastasis in patients with colon cancer,which is the most potential risk stratification method for liver metastasis. |
PDF Full Text Request