Detecting brain tumors is always an important but challenging task for MRI physicists and radiologists. Many automated brain tumor detection methods have been introduced. Due to the complexity and limitations of those procedures, practical applications of those methods in clinical setting are restricted. Based on a large sample size of acquired normal human datasets, this study provided a new approach to detect grey matter abnormalities in individual subjects. This approach consisted of three steps. First, voxel-based morphometry was performed on T1 images from 1000 normal subjects and each of the five tumor subjects. Second, we computed the distribution of the grey matter intensities of each voxel from 1000 normal subjects. Third, we compared the intensity of each grey matter voxel of tumor patient to its corresponding voxel's distribution of normal subjects. Finally, for each tumor subject, grey matter voxels that were out of 99.9% normal distribution were marked. This method uses large data sets to detect abnormalities and has the potential to be utilized in clinical applications.