Overview on Partial Volume Estimation in Brain MRI: Models and Methods
Jussi Tohka, Alex P. Zijdenbos, Ulla Ruotsalainen, Alan Charles Evans · 2003
Quantitative analysis of magnetic resonance (MR) images to gain knowledge about anatomy of human brain is increasingly important. For example, disorders or healthy aging can cause structural changes in the brain. These changes can be quantified by measuring properties of the anatomical structures of interest. However, it is not straight-forward to extract these structures of interest from images for quantification. For example, a single voxel may consists of several brain structures. This phenomenon, termed partial volume effect (PVE), is caused by the finite spatial resolution of imaging devices. Due to the complexity of human anatomy, the PVE is an important factor when an accurate structure extraction is needed. A standard segmentation problem within MRI is the task of labeling voxels according to their tissue type that are white matter (WM), gray matter (GM), and cerebro spinal fluid (CSF). An extension of this classification task is partial volume (PV) estimation, the estimation of the amount of each tissue type within each voxel. In this abstract, we present a concise overview of statistically-based PV estimation.