- Podcast
- 09/06/2026
- UK
Diagnosing Orthopaedic Infection By Identifying Neutrophils In Whole Histology Slide Images With Machine Learning Trained On Publicly Available Datasets
Description
In this episode of AI Talks with Bone & Joint, Simon and Amy discuss a March 2026 study by K Bentick and colleagues on using machine learning to diagnose orthopaedic infection by identifying neutrophils in whole histology slide images. The conversation focuses on a YOLO object detection model trained on publicly available datasets, including 3,923 images, to automate neutrophil counting in histology slides for detecting periprosthetic joint infection (PJI), a serious complication affecting about 1–2% of joint arthroplasty surgeries in the UK.
The hosts explain that the goal was to replace or support the manual, time-consuming, and error-prone process of identifying neutrophils in tissue samples. The model was evaluated on ground truth images and additional unseen cases, achieving strong performance with 82% precision, 79% recall, and an 80% F1 score. When compared with formal histopathological, microbiological, and multidisciplinary team diagnoses, the model remained comparably accurate, and for infection diagnosis it achieved 82% recall and an 86% F1 score versus MDT diagnosis. It correctly identified 9 of 10 infected cases and ruled out 7 of 9 non-infected cases.
The discussion highlights the potential clinical value of the model in helping pathologists by locating neutrophil hotspots and reducing workload, while keeping the human expert involved in final diagnosis. The speakers also note limitations, including the retrospective design and possible issues with generalizability across different laboratories, tissue processing methods, scanners, staining techniques, and image resolutions. They conclude that, despite these limitations, the study is an important step toward integrating machine learning into pathology and could improve efficiency and diagnostic accuracy in orthopaedic infection diagnosis.
Part of: Surgical Techniques and Training Collection
"Diagnosing Orthopaedic Infection By Identifying Neutrophils In Whole Histology Slide Images With Machine Learning Trained On Publicly Available Datasets" is included in the following Surgical Techniques and Training playlist: