Deep Learning-Based Open Source Toolkit for Eosinophil Detection in Pediatric Eosinophilic Esophagitis

GitHub Link

The GitHub link is https://github.com/hrlblab/open-eoe

Introduce

The Open-EoE project offers an open-source toolkit for detecting eosinophils (Eos) in whole slide images (WSIs) of Eosinophilic Esophagitis (EoE), a chronic esophageal disease. The toolkit supports three deep learning-based object detection models and implements ensemble learning for improved accuracy. It achieves a 91% accuracy in detecting Eos on WSIs, aligning well with pathologist evaluations, thus showing potential for integrating machine learning into EoE diagnosis. Installation and usage instructions are provided in the content. Additionally, a Docker image is available for easy implementation. Eosinophilic Esophagitis (EoE) is a chronic, immune/antigen-mediated esophageal disease, characterized by symptoms related to esophageal dysfunction and histological evidence of eosinophil-dominant inflammation.

Content

Eosinophilic Esophagitis (EoE) is a chronic, immune/antigen-mediated esophageal disease, characterized by symptoms related to esophageal dysfunction and histological evidence of eosinophil-dominant inflammation. Owing to the intricate microscopic representation of EoE in imaging, current methodologies which depend on manual identification are not only labor-intensive but also prone to inaccuracies. In this study, we develop an open-source toolkit, named Open-EoE, to perform end-to-end whole slide image (WSI) level eosinophil (Eos) detection using one line of command via Docker. Specifically, the toolkit supports three state-of-the-art deep learning-based object detection models. Furthermore, Open-EoE further optimizes the performance by implementing an ensemble learning strategy, and enhancing the precision and reliability of our results. The experimental results demonstrated that the Open-EoE toolkit can efficiently detect Eos on a testing set with 289 WSIs. At the widely accepted threshold of ³ 15 Eos per high power field (HPF) for diagnosing EoE, the Open-EoE achieved an accuracy of 91%, showing decent consistency with pathologist evaluations. This suggests a promising avenue for integrating machine learning methodologies into the diagnostic process for EoE. Please refer to INSTALL.md for installation instructions of the detection phase. you can put your all WSIs in a folder. if you want to use faster-rcnn as the model: if you want to use mask-rcnn as the model: if you want to use centernet as the model: if you want to use all of these three models: get the maximum Eos count in HPF First you need to put a folder include your data named WSIs in the container

---