Figure 4. (A) Original view and (B) view after applying the Positive
Pixel Count algorithm v9 of digitized images of an ORO-stained frozen
section at 40X magnification. The expert pathologist (AES) assessed
global HS and MaS based on the original view, and the algorithm
quantified the ORO stain percentage based on the stained slide pixel
area. The ORO-stained lipid droplets (positive pixels) are in orange (A)
in brightfield and in red (B) after the application of the Positive
Pixel Count algorithm v9.
Considering the better linear correlation of the Raman signal with the
TG results in the biochemical validation stage, compared to the
Reflectance measurements, the Raman signal was used as the primary
variable for predicting HS in the histopathological validation stage.
The boxplots in Figure 5 show the distribution of fat content
estimated by the Raman system, grouped by the degree of large droplet
MaS (assessed by the expert pathologist), ORO stain percentage
(estimated by the algorithm), and total steatosis percentage (evaluated
by the expert pathologist).