Computer aided iron quantification in digital images of liver biopsies.

Liver biopsies are an invaluable tool for the prognosis and diagnosis of iron overload disorders. Perls Stain (PS) is exploited to visualize iron accumulation in blue, quantified in our study in 10 whole-slide images (WSI) of liver biopsies. We developed a pipeline able to objectively estimate: percentage liver biopsy covered by PS, mean intensity of PS stain, density and dimension of PS granules, and PS heterogeneity. To reproduce the procedure followed by pathologists we performed our analysis at different zoom scales, while simultaneously applying image-wide color correction. We developed two methods: optical density color thresholding and a custom spectral phasor approach, both characterized by $<5%$ error when compared to pathologist's semi-quantitative evaluation. Furthermore, to improve intra-section evaluation, we generated heatmap overlays to highlight statistically significant tissue patches. Our pipeline paves the way toward a rapid and objective method for the evaluation of hepatic iron accumulation. Moreover, it can also be tuned and applied to different staining protocols ($e.g.$, PicroSirius Red for the quantification of liver fibrosis).