| Abstract: |
3-D histology has become an attractive technique providing insights into morphology of histologic specimens. However, existing techniques in generating 3-D views from a stack of whole slide images are scarce or suffer from poor co-registration performance when displaying diagnostically important areas at sub-cellular resolution. Our team developed a new scale-invariant feature transform (SIFT)-based workflow to co-register histology images and facilitate 3-D visualization of micro-structures important in histopathology of lung adenocarcinoma. The co-registration accuracy and visualization capacity of the workflow were tested by digitally perturbing the staining coloration seven times. The perturbation slightly affected the co-registration but overall the co-registration errors remained very small when compared to those published to date. The workflow yielded accurate visualizations of expert-selected regions permitting confident 3-D evaluation of the clusters. Our workflow could support the evaluation of histologically complex tumors such as lung adenocarcinomas that are currently routinely viewed by pathologists in 2-D on slides, but could benefit from 3-D visualization. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021. |
