Multi-modality Prostate Image Registration via Distance Maps

While prostate cancer is very common in men, it is usually indolent; most cases are slow growing and require minimal treatment. A method to noninvasively detect malignant cases is needed to deliver appropriate treatments when necessary; prostate cancer is the second leading cause of cancer death in American men. The current standard of care is needle biopsies, which are invasive and have a high false-negative rate.

 

Our collaborators at Duke University, led by Dr. Mark Palmeri and Dr. Kathryn Nightingale, are developing new methods to noninvasively image prostate cancer with ultrasound. The new imaging methods, acoustic radiation force imaging (ARFI) and shear wave elasticity imaging (SWEI), detect the increased stiffness associated with malignant lesions.

 

Improved non-invasive imaging assessment may be possible if we combine the results from multiple imaging methods, including magnetic resonance (MR)-based images. Solving the deformable image registration problem also allows use to validate our results against whole-mount histopathology images.

 

Our current approach is illustrated in the video below, created by Duke University student Tyler Glass:

 

 

Here we are using the visualization and analysis capabilities of 3D Slicer and ITK. The software and approach is based on the SlicerProstate extension, which performs registration of distance maps generated from whole-prostate segmentations. We extended the registration method to generate improved results by leveraging the topology of prostate structures. Since the prostate capsule, central gland, and urethra all encapsulate each other, a distance map generated from their combined segmentations provides more guidance to deformation inside the prostate. Initial results are promising, and we look forward to improving and disseminating the approach.

This work was supported by the National Cancer Institute of the National Institutes of Health under Award Number R41CA196565.

Questions or comments are always welcome!