High resolution 3-D MR image reconstruction from multiple views
Prasenjit Mondal, Jayanta Mukherjee, Shamik Sural, Pinak Pani Bhattacharyya · 2012
This paper introduces an efficient technique to reconstruct the volume of human brain from low resolution MR image sequences acquired in sagittal, coronal and axial views. The algorithm uses image interpolation as the basis of volume reconstruction. It works in different stages. At first, the brightness and contrast of the MR images in different views are normalized based on the repetition time (TR) and echo time (TE) used during the MR imaging. Alignment of MR image sequences in different views is performed based on a standard 3D human brain anatomical model. Based on the alignment information, the spatial resolution of the MR images in all of the three views are normalized. Furthermore, a technique using normalized cross-correlation has been applied to refine the alignment process. Finally, the high resolution volume of the brain is reconstructed by interpolation based on inverse distance weighing. The proposed volume reconstruction scheme shows promising results producing good quality of the reconstructed volume images.