Measuring ocular toxicologic changes with pattern recognition

The eye serves as an important organ for toxicology research. Below is a rodent H&E ocular cross-section, followed by pathologist-trained pattern recognition of each layer. This provides a larger and faster region of interest drawing than would be used with manual ROI drawing by a technician or pathologist. Depending on the performance, Flagship will use pattern recognition in some cases and manual technician drawing with pathologist supervision as shown below.

Cross-section of a rodent retina stained with H&E

Genie analysis markup across a retina section
Histology pattern recognition teaches the computer to recognize different layers of the retina. Flagship uses this technique to then quantitate morphologic changes in each layer

In the example above, there is some error in the automated pattern recognition approach for estabilishing regions of interest. The Flagship pathologist assesses the performance on each slide, and will determine if the accuracy is acceptable, or if manual region of interest drawing is required as shown below.

Manual region of interest drawing in retina sections is sometimes preferable to running histology pattern recognition tools like Aperio's Genie
If the performance of histology pattern recognition is not accurate enough in a given slide, the Flagship pathologist will choose to draw the tissue sections manually. This is frequently more accurate, but also more time-consuming.

Once Flagship identifies the cell layers, either manually or with histology pattern recognition (or a mixture of both), cell and area measurements can be combined to give quantitative changes of cell layer densities or other layer abnormalities between control and treated animals. After either manual ROI or histology pattern recognition, cell counting and cell density can be determined on the inner and outer nuclear layers, or the ganglion cell layer, for example.

Digital pathology nuclear measurements in outer and inner nuclear layers of the retina
After either manual ROI or histology pattern recognition, cell counting and cell density can be determined on the inner and outer nuclear layers
Genie and digital pathology used to identify only the ganglion cell layer after using Aperio's Genie and pattern recognition in ocular toxicologic pathology
With pattern recognition, we can look at only one layer (e.g. the ganglion cell layer) for further image analysis
Ophthalmology analysis of cell layer thickness in the retina with Aperio ImageScope measurement tools
Morphometric measurements like cell layer thickness can be used to compare treated with control groups.