University of Minnesota
University Relations
myU OneStop

Go to unit's home.

Home | DTI | 2007–08 funded proposals | Gregory J. Metzger, Alexander M. Truskinovsky, Kenneth S. Koenemann

Initiatives in Digital Technology: 2007–08 Funded Proposals

Gregory J. Metzger, Alexander M. Truskinovsky, Kenneth S. Koenemann

Construction of Three-Dimensional Co-Registered Pathology Data as a Gold Standard for the Classification of Magnetic Resonance Imaging of Prostate Cancer

The full potential of magnetic resonance imaging to improve the diagnosis and management of prostate cancer has not yet been fully realized. MRI allows the non-invasive acquisition of three-dimensional (3D) anatomic, vascular and metabolic information. While MRI is currently being used clinically to aid in diagnosis and staging, it has the potential to provide the necessary information to target focal therapies and develop patient specific treatment strategies. However, in order to realize this potential, more rigorous validation studies must be undertaken to understand the best way to use the multi-parametric MRI to determine the extent (location and volume) of cancer and its aggressiveness. Pathology data acquired after the removal of the prostate is the best gold standard to use for validation purposes. Detailed analysis of pathology slides made from the removed prostate can be directly compared with MRI data acquired prior to surgery. However, before this comparison can be made, the interpreted pathology slides must be reconstructed back into a 3D volume and co-registered to the in vivo MRI data. This is a non-trivial task as the MRI data is acquired with an endorectal coil which locally distorts the prostate in vivo, while ex vivo, the prostate also changes shape.This project will focus on developing the tools necessary to reconstruct the 3D volume from multiple digitized pathology slides and registration of that volume to MRI. The methods developed in this work will make it possible to use pathology results as a gold standard for training statistical classifiers based on the MRI data. It is envisioned that these classifiers will be able to generate 3D probability maps of cancer with the MRI data alone. Current collaborations between radiology, pathology and urology are already producing the data required as input for the methods proposed in this project.