Susan M. Noworolski1,2, Richard F. Guo3, Galen D. Reed1, Kyle Kuchinsky4, Kirsten Greene5, Peter Carroll5, Daniel B. Vigneron1,2, John Kurhanewicz1,2, Jeffry Simko4
1Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, USA; 2Graduate Group in Bioengineering, University of California, San Francisco & Berkeley, San Francisco & Berkeley, CA, USA; 3Electrical Engineering, University of California, Berkeley, Berkeley, CA, USA; 4Pathology, University of California, San Francisco, San Francisco, CA, USA; 5Urology, University of California, San Francisco, San Francisco, CA, USA
Correlation of MR images to histopathology, the gold-standard for tissue identification, is important for evaluating MR's role in diagnosis. Digitized, whole-mount histopathology slides of the prostates of ten patients were aligned to MR images. The %overlap with only rigid alignment was 656% which increased to 990.6% after automatic deformation. Visually identified landmarks were 2.51.5 mm distant on average, but ranged as far as 8mm. Non-rigid alignment can greatly improve correlation of MR data to histopathology but nonuniform deformation of the prostate may remain and needs to be considered before using the aligned histopathology to classify tissue types on MRI.