Digital Hepatic Iron Content: An Artificial Intelligence Model for Spatially-Resolved Histologic Iron Quantitative Analysis in Liver Samples.
Researchers
Journal
Modalities
Models
Abstract
Currently, precise evaluation of tissue hepatic iron content (HIC) requires laboratory testing by tissue-destructive methods based on colorimetry or spectrophotometry. In order to maximize the use of routine histologic stains in this context, we developed an AI model for recognition and spatially-resolved measurement of iron in liver samples. Our artificial intelligence (AI) model was developed using a cloud-based, supervised deep learning platform (Aiforia Technologies, Helsinki, Finland). Using digitized Pearl’s Prussian Blue iron stain WSIs representing the full spectrum of changes seen in hepatic iron overload, our training set consisted of 59 cases, and our validation set of 19 cases. The study group consisted of 98 liver samples from 5 different laboratories, for which tissue quantitative analysis by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) was available, collected between 2012 and 2022. The correlation between the AI model % iron area and hepatic iron content (HIC) was Rs=0.93 for needle core biopsy samples (n=73) and Rs=0.86 for all samples (n=98). The digital hepatic iron index (dHII) was highly correlated with hepatic iron index (HII)>1 (AUC=0.93) and HII>1.9 (AUC=0.94). The percentage area of iron within hepatocytes (versus Kupffer cells and portal tract iron) identified patients with any HH-related mutations (either homozygous or heterozygous) (AUC=0.65, P=0.01) with at least similar accuracy than HIC, HII, and any histologic iron score. The correlation between the Deugnier and Turlin score and the AI model % iron area for all patients was Rs=0.87 for total score, Rs=0.82 for hepatocyte iron, and Rs=0.84 for Kupffer cell iron. Iron quantitative analysis by our AI model was highly correlated to both detailed histologic scoring systems and tissue quantitative analysis by ICP-MS and offers advantages (related to the spatial resolution of iron analysis and the non-tissue destructive nature of the test) over standard quantitative methods.Copyright © 2023. Published by Elsevier Inc.