| Cancer is one of major killers harmful to human health in which the intracranial tumor is a common disease, particularly glioma, which features infiltrative growth, boundary not clear, high recurrence rate after surgery, routine chemotherapy and radiotherapy big side-effected and ineffective. Generally higher level of pathological grade of glioma, the higher the degree of malignancy, thus the worse the prognosis. Currently, the pathological grade of glioma plays an important role in the choice of treatment program, therefore, the correct diagnosis of the pathological grade has important significance in clinic.This paper focuses on the pathological grade of glioma based on magnetic resonance images. First of all, we study on the format and analytical method of DICOM file, retrieve the relevant information from the MR images and pretreat it to be anonymous; Segmentation of ROI was performed based on region growing combined with morphology. After establishing the measuring method for surface area, volume and other morphological parameters of glioma, we developed the software using object-oriented analysis and design method. In the end, ROI extraction, measurement and save are executived for 94 glioma patients. Meanwhile, the quantitative analysis of the characteristic parameters was performed using statistical software SPSS 17.0 according to independent sample Kruskal-Wallis H test and independent sample T test. A preliminary analysis showed that the higher level of glioma, the bigger tumor surface area and volume, therefore, surface area and volume can be used to identify grade of glioma. These results not only further the clinical pathological grade of glioma based on MRI, but also provide guidance and direct technical support for classification and diagnosis of other cancer. |
