Background
To characterise cancer tissue, it is often necessary to conduct multimodal imaging and correlate the resulting datasets with histopathology. To do this accurately can be challenging, especially the correlation of MRI at different field strengths with histopathology, which requires the specimen to be sliced and photographed. This is of relevance for our project “Correlation of Ultra-high Field MRI with Histopathology for Evaluation of Rectal Cancer Heterogeneity” because the histopathology is the ground truth for the training of machine learning models for the MR images to be classified into regions of cancer, desmoplasia and fibrosis.
Aim
To develop a reliable and robust technique for the accurate and precise correlation of preclinical MRI datasets with histopathology.
Methods
The method we have developed uses a variety of designs of a neckless 3D printed vial (a ‘Fiducial Vial’) with grooves running differential lengths down its internal sides. The specimen is embedded in 4% agarose gel within the vial. The gel is visible in the MR image, with the addition of a small amount of Magnevist serving to differentiate it from the specimen by altering its contrast. An axial MRI cross section through the vial reveals both the tissue morphology and its position within the tube by the number of grooves visible on the interior of the vial. A removable base (a little like a cake tin) enables the embedded specimen and gel to be easily removed intact by pushing them out from the bottom.
Results
Previous MRI scans in our study used scintillation vials or falcon tubes. The fiducial vial provides a geometric co-ordinate system so that the exact location of any part of tissue can be located on the MRI, and subsequently correlated with histopathology. The adoption of the 3D printed vial does not change the scanning protocol. After scanning, the gel and specimen can be sliced into sections. The gel profile enables the axial position of the MRI slice to be easily determined. Since the gel is 4% agarose, for those samples where a cryostat can be used, the gel just needs to be cooled prior to slicing.
Conclusion
Previous work on this project used painstaking methods to co-register the MRI and histopathology without a marker system. The method described in this abstract is significantly faster and more precise. We are applying this and related methods to all specimens in our study.