Computer aided diagnosis workstation for brain tumor assessment

The paper presents a Computer Aided Diagnosis (CAD) software for brain tumor detection and analysis from Magnetic Resonance Imaging (MRI). The software utilizes a novel multi-stage method that is capable of dealing with three main types of brain tumors, i.e. HG gliomas, metastases and meningiomas and yields object masks as well as quantitative parameters. The processing method makes use of several available in MRI series: FLAIR, T1-Weighted, Contrast Enhanced T1-Weighted and Perfusion Weighted Images. Image processing methods involve registration of MR series, Region of Interest determination, multi-stage image segmentation, and finally analysis of neovasculature in suspected areas basing on perfusion maps. The brain regions are then labeled as normal, tumor or peritumoral. Relative perfusion coefficient is calculated with respect to a healthy white matter of the brain. Obtained results are presented to the radiologist. The presented CAD workstation is able to communicate with PACS (Picture Archiving and Communication System) using DICOM protocol.