School of Computing Research Colloquia

Multi-modal image registration: from histopathology to MR images

Yi Song, Computer Vision research group

The histological examination of adjacent tissue sections prepared with different stains or biomarkers can provide a valuable amount of information to help understanding of the physical or functional properties of tissue at microscopic level. MR imaging is often used for non-invasive tissue characterization at macro scale. Spatial correspondence between histology and MR images allows a complete picture of the relationship between the causative disease and the resulting tissue structure to be built up. However, shrinkage and deformation occurring during the excision of the tissue specimen and the histological processing complicate the alignment between histological sections and their co-registration to MR images.

Our research is to develop efficient multi-modal image registration methods which address the challenges of 1) alignment between histological sections with different stains, 2) co-registration histological sections with MR images. This work is funded by Wellcome Trust and the EPSRC, as part of the £11 million project of "fifty activate years after fifty".