Differential Diagnosis Of Pancreatic Cancer Based On MRI Image

Pancreatic cancer is one kind of tumor that has very high degree of malignant and it is one of the highest death ratecancer all over the world,many effort have been put in to research the prevention, diagnosis, prognosis and treatment of pancreatic cancer. CT has a relatively good performance in the diagnosis, surgical resectability assessment andstaging of pancreatic cancer, but when the lesions is small and the lesions is not obvious, thediagnosis effect of CT is not satisfactory. MRI is a rapidly developing imaging techniques in recent years, it showed a better results in the diagnosis of pancreatic cancer, many researchers believe that MRI is superior to CT in the diagnosis of pancreatic cancer.As one of thecurrentprimary means ofdiagnosisof pancreatic,MRIshowed alot of advantages, especially combined with DWI and MRCP technique,the diagnosis effect beyond other diagnostic tools.MRI and CT technology astwo most common imaging techniquein the diagnosis of pancreatic cancer, manyresearchers have compared the performance of diagnosis of this two techniques,the resultsobtainedmany inconsistencies, or even completely oppositeconclusion, we made a meta-analysis by integrating the related research in recent years to make a new unified conclusion, to provide guidance for the diagnosis of pancreatic cancer. In this meta-analysis, we completed the followingwork. First, we search the literature over the past decade about the performance of MRI and CT in the diagnosis of pancreatic cancer,screen thepotentialrelevant literature strictly accordance with the inclusion criteria, and finally brought into some high quality documents for subsequent research; secondly, according to data select and extract criteria to extract the necessary data required for this study and complete the meta analysis; finally, we did the subgroup analysis on the factors that may affect the diagnosis performance of MRI and CT,and calculated the significant degree by the statistical test.Through the results we can find out that CT and MRI have equal sensitivities and that MRI provides high specificity in differentiating between benign and malignant pancreatic tumors. These findings suggest that MRI may be a better choice for diagnosing pancreatic carcinoma.Despite imaging technique brought great convenience for the diagnosis of pancreatic cancer, but to diagnose pancreatic canceraccurately remains a problem, in order to assist thedoctorto diagnose pancreatic cancer and reduce the work of readingimage,we completed the study of computer-aided diagnosis based on the image of MR, and compared the classfication accuracy of SVM and LDA method to classfy the pepoleof pancreatic cancer, pancreatitis patients and normal. Themain work consists following processes: we collect data in hospital, extract the image featureunder the guidanceof experts, achieve themachine learning algorithms to train the training data set, and then quantify the pancreatic diseasefeature,to convert fromthe human visual characteristic feature to the computer data set, finally, classfying the patients of pancreatic diseases, to offerthe service of computer aided diagnosis for the clinical.