Strategy for removal of surrounding noise in 3D head tomographic images without changing brain image data

For image-quided neurosurgery, the 3D reconstruction of CT/MR images must faithefully retain spatial and tissue resolution. A strategy for removal of surrounding noise in practical 3D visualization of head CT and MRI Images without changing brain image data is represented in this paper. In this strategy average gray-level values of surrounding noise and head image data, Including the histogram of brain image data, are used as adaptive parameters to construct a 2D ellipse Gaussian curved thresholding surface for automatic segmentation of surrounding noise according to geometric features of the head CT/MRI scans. Then a recursive labeling method is used to remove the surrounding noise-spots and noise-blocks. Unlike common noise reducing methods, this strategy does not change brain image data so that original image resolution and position can be retained.