Electronic Thesis and Dissertation Repository


Master of Science


Medical Biophysics


Dr. Charles McKenzie


Novel magnetic resonance (MR) imaging techniques can be validated using accurate co-registration with histology. Whole-animal histological sections allow for simultaneous analysis of multiple tissues, and may also aid in registration by providing contextual information and structural support to tissues which if isolated from the body would be difficult to register.

This thesis explores the feasibility of co-registration between whole mouse histology with 3D MR images using an intermediate optical image volume acquired during tissue sectioning. Of the two transformations required for this approach, 3D co-registration of MR and optical images is more challenging to perform due to changes in contrast, slice orientation, and resolution between these modalities. Here, an automated non-rigid registration technique utilizing mutual information is proposed to accurately register 3D whole mouse optical and MR images as a first step towards automated registration of histology. Validation of this technique was accomplished through calculation of post-registration target registration error.