Efficient construction of histology slide mosaics via phase correlation registration of high resolution tiles

During development of new and experimental imaging techniques, one must compare with other well established methods to qualify the validity of the new methods. The use of diffusion tensor magnetic resonance imaging for the visualization of collateral nerve fibers in the spinal cord was recently proposed. To validate this new imaging method, the MR results need to be compared with histology tissue samples in which the nerve fibers are clarified through staining. We present here a method for the mostly-automatic merging of very high resolution “tiles” of the histology slides to construct a composite image of a tissue slice. This application is novel in that it uses a phase correlation registration techniques with sub-pixel resolution to construct the composite image. Registration error detection is achieved by comparing the sub-pixel registration result with a standard phase correlation estimate.