Diagnosis of Inflammation from H&E Stained Histology Images

Tech ID: 15-052

Inventors: Dr. Danny Chen, Dr. Jiazhuo Wang, Dr. John MacKenzies, Dr. Raga Ramachandran

Date added: August 30, 2020


A novel method for automatically identifying and quantifying neutrophils in H&E stained histology tissue images. 

Technology Summary

As inflammatory diseases such as inflammatory bowel disease and COPD become more prevalent health issues, improving diagnostic capabilities is essential to elevating the standard of care. Identifying neutrophils in hematoxylin and eosin (H&E) stained tissue images is a key step in diagnosing acute inflammation, but is currently done manually.  Unfortunately, certain structural characteristics of neutrophils (eg - multiple lobes per cell nucleus and a cytoplasm that is hard to see), make them difficult to identify and quantify. As a result, accurate segmentation of these subcellular structures varies by pathologists and is a time-consuming process, one that could benefit from automation and identification of clear biomarkers. 

Imgonline Com Ua Resize Lbyrmpmtcxzw

Researchers at the University of Notre Dame have recently developed an automated solution for inflammatory disease diagnosis that improves by 80% the confidence that pathologists have in identifying and counting neutrophils. The method makes use of an image obtained in H&E stained tissue samples commonly found in pathology labs and uses computer analysis to separate and identify white blood cell indicators of inflammatory disease. The algorithm distinguishes hard-to-see cells and provides indicators to direct the pathologist toward slides that should receive extra examination. The Diagnosis of Inflammation from H&E Stained Images technology developed at Notre Dame is quicker, more accurate, and more efficient in diagnosing inflammatory diseases. 

Market Advantages

•    Time Savings: Highlighting of neutrophils reduces time spent by pathologists manually counting these biomarkers for each viewed tissue image.
•    Increased Quality: identifies ~90% of present neutrophils in tissue images, as compared to pathologists who report that they are confident in manually identifying only ~50% of present neutrophils.

Market Opportunity

•    Licensed feature for digital pathology software - $22M Market
•    Cancer immunotherapy research - $1.2M Market


Identifying Neutrophils in H&E Staining Histology Tissue Images.

doi: 10.1007/978-3-319-10404-1_10


Segmenting Subcellular Structures in Histology Tissue Images.

doi: 10.1109/ISBI.2015.7163934

Intellectual Property

US Patent 10,121,245

Technology Readiness Status

TRL 3 - Experimental Proof of Concept


Richard Cox