Aligning prostate MRI with histopathology is difficult because the two are created in completely different ways. MRI captures the prostate in a consistent 3D orientation, while histopathology involves removing the prostate, sectioning it into thin slices, and examining those tissue sections under a microscope. During this process, the specimen can deform and the slices do not naturally line up with the original imaging. This makes it harder to accurately map where cancer identified on pathology exists on MRI. In this work, the RAPHIA pipeline was developed to improve that alignment by automating key parts of the process, reducing manual input while improving both speed and accuracy. This makes it more practical to generate large datasets where cancer confirmed on pathology can be mapped back onto MRI for AI development, while also creating a feedback loop where imaging can be compared directly to pathology to review and refine lesion grading.

The 3DQ Lab contributed by addressing the problem at the source, how the specimen is cut. From 2018 to 2024, approximately 720 patient specific slicing guides were created from 3D models of each prostate and 3D printed using FDM, then provided directly to pathology. These guides helped ensure that tissue sections were cut in a way that better matched the original MRI orientation, reducing variability introduced during specimen preparation. The designs were iterated in collaboration with the Department of Urology, with adjustments to dimensions and spacing to improve alignment and usability. By making the physical slices more consistent with the imaging data, the guides supported a more reliable downstream alignment process and helped enable more accurate comparison between MRI and pathology.

Publication Link: ScienceDirect