Pathologist-Trained Machine Learning Classifiers Quantitate Celiac Disease Features | May 30

Learn how HALO AI is advancing understanding of Celiac disease

Pathologist-Trained Machine Learning Classifiers Quantitate Celiac Disease Features

Date: 30 May 2024
Time: 8:00 – 9:00 PDT | 11:00 – 12:00 EDT | 16:00 – 17:00 BST
Location: Webinar



Histologic evaluation of the mucosal changes associated with celiac disease is important for establishing an accurate diagnosis and monitoring the impact of investigational therapies. While the Marsh-Oberhuber classification has been used to categorize the histologic findings into discrete stages (i.e. type 0-3c), significant variability has been documented between observers using this ordinal scoring system. Therefore, a pathologist-trained machine learning classifier was developed to objectively quantitate the pathological changes of villus blunting, intraepithelial lymphocytosis, and crypt hyperplasia in small intestine endoscopic biopsies.

A HALO AI DenseNet2 convolutional neural network (Indica Labs) was trained and combined with a second HALO algorithm to quantitate intraepithelial lymphocytes on CD3 immunohistochemistry whole slide images and used to correlate feature outputs with ground truth modified Marsh scores in 116 samples. Median %CD3 counts (positive cells/enterocytes) from villous epithelium increased with higher Marsh scores. Indicators of villus blunting and crypt hyperplasia were also observed. Using these individual features, a combined feature machine learning score (MLS) was created to evaluate a set of 28 matched pre- and post-intervention biopsies captured before and after dietary gluten restriction. The disposition of the continuous MLS paired biopsy result aligned with the Marsh score in 96.4% (27/28) of the cohort. Machine learning classifiers can objectively quantify histologic features and capture additional data not achievable with manual scoring. Such approaches should be further investigated to improve biopsy evaluation, especially for clinical trials.

Learning Objectives

  • Understand the prevalence and impact of celiac disease among patients and appreciate manual methods for celiac disease scoring and their associated limitations
  • Discover how a HALO AI network was trained to characterize histopathologic features of celiac disease from IHC-stained tissues
  • Learn how machine learning-assisted pathologist evaluations can reduce intra/inter-observer variability and provide quantitative data that improve upon traditional categorical scoring systems
  • Gain insight on how machine learning classifiers can enhance endpoint assessment in clinical trials of investigational therapies


Kelly Credille, DVM, PhD, Dip ACVP
Clinical Diagnostics Laboratory, Eli Lilly and Co.

Kelly has focused her career on comparative pathology and translational medicine to improve human and animal health. In her current role in the Clinical Diagnostics Lab within Lilly Research Laboratories, she develops biomarkers by aligning gene and protein expression with morphology in order to match patients to medicines. As a consequence she has a strong interest in image analysis and the application of deep learning tools to morphology and pathology. To this end, she believes the future of anatomic pathology is its transition from a descriptive to a quantitative discipline. Kelly also supported safety assessment of new medicines in Toxicology for several years. Prior to Lilly, in academia, she achieved internationally recognized expertise in diagnostic and research comparative dermatopathology. Kelly earned her DVM and PhD at Michigan State University and trained in anatomic pathology at Cornell University, receiving her certification from the American College of Veterinary Pathologists.

Share This Post