To detect lesions of brain CT automatically, a statistical atlas of attribute vectors (SAAV) was designed and created to describe the multiple features of medical images. By comparing the features of study image with those of SAAV, we successfully detected various kinds of brain lesion. It was demonstrated that the algorithm is effective in detecting various kinds of lesions found on brain CT images. Further studies are needed to make the algorithm more acceptable.
Algorithms ; Brain Diseases ; diagnostic imaging ; Humans ; Image Processing, Computer-Assisted ; Numerical Analysis, Computer-Assisted ; Tomography, X-Ray Computed