Automatic Classification Of Brain Tumor Based On In Vivo Multi-Voxels Mr Spectroscopy

The brain tumor is a kind of common disease in nervous system, which poses a serious threat to life with its high disability and fatal rate. Brain tumor patients without proper treatments have an average surviving time of 1-2 years, which implies the importance of classifying and grading tumor for following clinical plans. The biopsy, Golden Standard for diagnosis, can provide accurate information while influenced by two factors:the location of the tumor and the physical conditions of the patients. Therefore, a non-invasive classification and grading of brain tumor would benefit patients a lot, especially those in need of pathological section to guide the following treatments. Magnetic Resonance spectroscopy (MRS) can quantitatively analyze human body’s metabolism, biochemical environment and compounds. These quantitate differences between different tissues can help establish a classify model to predict the type and grade of brain tumor, non-invasively. There have been many researches on the classification or grading of brain tumor, but barely no work in a comprehensive system combine classification and grading together. Therefore, we designed an automatic multi-layer classification system to simulate the brain tumor diagnosis process according to clinic workflow:1).To identify a brain tumor according to quantitative MRS parameters, such as Cho and NAA; 2).To diagnose the brain tumor type based on T1W/T2W MRI, MRS and clinical information. A tumor model was established and tumor with the highest score was the recognized type; 3).To indicate the grade of brain tumor based on T1W/T2W MRI and MRS. Support Vector Machine (SVM) was used in the Step 1 and 3, and a Tumor Model was established in Step 2. The whole three steps together constituted an automatic multi-layer classification system.This study proposed a novel multi-layer classification and grading system of brain tumor. It takes in many features that were easily ignored, and has reached an overall accuracy of 90%. It could be a helpful auxiliary tool for neurosurgeon and neuroradiologists. The system will develop along with the research work, and provide more effective support.