Related Experiment Video
Updated: Jul 23, 2026
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Using Artificial Intelligence to Stratify Normal versus Abnormal Chest X-rays: External Validation of a Deep Learning
Sarah R Blake1, Neelanjan Das1, Manoj Tadepalli2
1East Kent Hospitals University NHS Foundation Trust, Ashford TN24 OLZ, UK.
Diagnostics (Basel, Switzerland)
|November 24, 2023
Summary
Artificial intelligence (AI) software accurately classified chest radiographs (CXRs) as normal or abnormal. This AI tool has the potential to reduce reporting backlogs and improve patient care through earlier interventions.
Area of Science:
- Radiology
- Artificial Intelligence
- Medical Imaging
Background:
- Chest radiography (CXR) is the most common radiological exam globally.
- Increasing CXR volumes lead to reporting delays, impacting patient safety and timely treatment.
- Computer-aided detection (CAD) AI offers a solution for efficient and accurate CXR interpretation.
Purpose of the Study:
- To evaluate the performance of qXR, a CE-marked CAD software, in classifying normal versus abnormal CXRs.
- To assess the potential of AI in optimizing radiologist workflow and resource allocation.
- To determine if AI CXR classification can aid in managing reporting backlogs.
Main Methods:
- Retrospective cross-sectional study of 1040 CXRs from diverse clinical settings.
- Utilized qXR, an AI CAD software trained on over 4.4 million CXRs.
- Established ground truth by inter-radiologist agreement between two senior radiologists.
Main Results:
- The AI software achieved a sensitivity of 99.7% and a specificity of 67.4% in classifying CXRs.
- Performance remained consistent across various healthcare settings, patient demographics, and X-ray equipment.
- No statistically significant performance variations were observed in subgroup analyses.
Conclusions:
- The qXR AI software accurately distinguishes normal from abnormal CXRs.
- AI-powered classification can potentially alleviate reporting backlogs in radiology departments.
- Early patient intervention facilitated by AI may lead to improved clinical outcomes.
