Related Experiment Video
Updated: May 5, 2026
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
1.7K
Post-deployment performance of a deep learning algorithm for normal and abnormal chest X-ray classification: A study
Amina Abdelqadir Mohamed AlJasmi1, Hatem Ghonim2, Mohyi Eldin Fahmy2
1Emirates Health Services, DSO Digital Park Building A8, Dubai Silicon Oasis, Dubai, UAE.
European Journal of Radiology Open
|November 7, 2024
Summary
Artificial intelligence (AI) effectively screens chest radiographs (CXRs) for infectious diseases, demonstrating high accuracy in identifying normal cases. This AI integration significantly reduces reporting time and enhances diagnostic precision for radiologists.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence
Background:
- Chest radiographs (CXRs) are crucial for screening infectious diseases like tuberculosis and COVID-19 in high-volume settings.
- Manual CXR interpretation is time-consuming, necessitating efficient solutions for radiologists.
Purpose of the Study:
- To evaluate the real-world performance of an AI algorithm (qXR v2.1) in classifying chest radiographs during visa screening.
- To assess the AI's agreement with radiologist interpretations and gather feedback on implementation challenges and impact.
Main Methods:
- A post-deployment study analyzed 1,309,443 CXRs from January 2021 to June 2022 in the UAE.
- The qXR v2.1 software classified scans as normal or abnormal, with performance evaluated against radiologist consensus.
- A digital survey assessed healthcare professionals' perceptions of AI integration.
Main Results:
- The AI demonstrated a Negative Predictive Value (NPV) of 99.92% and an overall agreement of 72.90% with radiologists.
- Positive Predictive Value (PPV) was 5.06%.
- 88.2% of radiologists reported reduced turnaround times, and 82% noted improved diagnostic accuracy.
Conclusions:
- The AI algorithm reliably identifies normal CXRs with high NPV, making it suitable for large-scale screening.
- AI integration in CXR interpretation workflows can enhance efficiency and diagnostic accuracy in clinical practice.
