Fully automated histological classification of cell types and tissue regions of celiac disease is feasible and correlates with the Marsh score

AimsHistological assessment is essential for the diagnosis and management of celiac disease. Current scoring systems, including modified Marsh (Marsh-Oberhuber) score, lack inter-pathologist agreement. To address this unmet need, we aimed to develop a fully automated, quantitative approach for histology characterisation of celiac disease. MethodsConvolutional neural network models were trained using pathologist annotations of haematoxylin and eosin-stained biopsies of celiac disease mucosa and normal duodenum to identify cells, tissue and artifact regions. Human interpretable features were extracted and the strength of their correlation with Marsh scores were calculated using Spearman rank correlations. ResultsOur model accurately identified cells, tissue regions and artifacts, including distinguishing intraepithelial lymphocytes and differentiating villous epithelium from crypt epithelium. Proportional area measurements representing villous atrophy negatively correlated with Marsh scores (r=-0.79), while measurements indicative of crypt hyperplasia and intraepithelial lymphocytosis positively correlated (r=0.71 and r=0.44, respectively). Furthermore, features distinguishing celiac disease from normal colon were identified. ConclusionsOur novel model provides an explainable and fully automated approach for histology characterisation of celiac disease that correlates with modified Marsh scores, facilitating diagnosis, prognosis, clinical trials and treatment response monitoring. KEY MESSAGESO_ST_ABSWhat is already known on this topicC_ST_ABS[tpltrtarr] Prior research has utilised machine learning (ML) techniques to detect celiac disease and evaluate disease severity based on Marsh scores. [tpltrtarr]However, existing approaches lack the capability to provide fully explainable tissue segmentation and cell classifications across whole slide images in celiac disease histology. [tpltrtarr]The need for a more comprehensive and interpretable ML-based method for celiac disease diagnosis and characterisation is evident from the limitations of currently available scoring systems as well as inter-pathologist variability. What this study adds[tpltrtarr] This study is the first to introduce an explainable ML-based approach that provides comprehensive, objective celiac disease histology characterisation, overcoming inter-observer variability and offering a scalable tool for assessing disease severity and monitoring treatment response. How this study might affect research, practice or policy[tpltrtarr] This studys fully automated and ML-based histological analysis, including the correlation of Marsh scores, has the potential to enable more precise disease severity measurement, risk assessment and clinical trial endpoint evaluation, ultimately improving patient care.