Research On Cerebral Glioma Grading Andcompensatory Mechanism Based On BOLD-fMRI

Cerebral glioma, the most common and devastating cerebral tumor, remains to be lack of efficient treatments. With diffuse growth characteristic and difference between lesion location and affected cognitive function, glioma has brought challenges to make efficient surgery plan and protect brain cognitive function to the limits simultaneously.BOLD-f MRI is a noninvasive and high spatiotemporal resolution imaging technique, which has been a useful tool in brain functional research. What’s more, RS-f MRI requires no intricately experimental design and participants’ task cooperation, is more acceptable and suitable for clinical application.Our research focuses on the glioma grading and functional compensation to provide more information for the glioma diagnosis and treatment. Further, supplementary function in surgery planning will be offered.The main contents and innovations of the paper are as follows:1. Glioma grading based on SVM. We extract the four different features of RS-f MRI, including signal intensity difference ratio(SIDR), signal intensity correlation(SIC), fractional amplitude of low-frequency fluctuation(f ALFF) and regional homogeneity(Re Ho), and utilize Mann–Whitney U test to analyze the correlation between tumor grades and the extracted parameters. In addition, we evaluate the accuracy, sensitivity and specificity of these four parameters in tumor grading through SVM. The result indicates that SIC and f ALFF have significant difference between high grade glioma(HGG) and low grade glioma(LGG).The accuracy, sensitivity and specificity of SVM classification were better than 50%, where SIC had the best classification accuracy(89%).2. Compensatory mechanism of frontal glioma based on resting state brain network. Firstly, we establish brain network based on left frontal gliomas, right frontal gliomas and healthy subjects, and calculates the K and BC parameters of the whole brain, left brain and right brain. Then we compute the hubs of each network and analyze the hub difference among the networks. Lastly, we make correlation analysis between brain network parameters and MMSE grade. This result indicates that K and BC value of contralateral normal side are much higher than that of focus side. Meanwhile, the hubs trend to transform to the contralateral normal side. Cognitive ability is correlated to the network parameters significantly from the correlation analysis.