AI Assisted Detection of Chest X-Rays

Pleural Effusion Clinical Trial

— AID-CXR  

Official title:

Utility of an AI-based CXR Interpretation Tool in Assisting Diagnostic Accuracy, Speed, and Confidence of Healthcare Professionals: a Study Using 500 Retrospectively Collected Inpatient and Emergency Department CXRs From Two UK Hospital Trusts

Clinical Trial Details — Status: Active, not recruiting

Administrative data

• NCT number: NCT06075836
• Other study ID #: 310995 - B
• Status: Active, not recruiting
• Start date: October 31, 2023
• Est. completion date: June 2025

Study information

• Verified date: June 2024
• Source: Oxford University Hospitals NHS Trust
• Contact: n/a
• Is FDA regulated: No
• Study type: Observational

Clinical Trial Summary

This study has been added as a sub study to the Simulation Training for Emergency Department Imaging 2 study (ClinicalTrials.gov ID NCT05427838).

The Lunit INSIGHT CXR is a validation study that aims to assess the utility of an Artificial Intelligence-based (AI) chest X-ray (CXR) interpretation tool in assisting the diagnostic accuracy, speed, and confidence of a varied group of healthcare professionals. The study will be conducted using 500 retrospectively collected inpatient and emergency department CXRs from two United Kingdom (UK) hospital trusts. Two fellowship trained thoracic radiologists will independently review all studies to establish the ground truth reference standard. The Lunit INSIGHT CXR tool will be used to analyze each CXR, and its performance will be measured against the expert readers. The study will evaluate the utility of the algorithm in improving reader accuracy and confidence as measured by sensitivity, specificity, positive predictive value, and negative predictive value. The study will measure the performance of the algorithm against ten abnormal findings, including pulmonary nodules/mass, consolidation, pneumothorax, atelectasis, calcification, cardiomegaly, fibrosis, mediastinal widening, pleural effusion, and pneumoperitoneum. The study will involve readers from various clinical professional groups with and without the assistance of Lunit INSIGHT CXR. The study will provide evidence on the impact of AI algorithms in assisting healthcare professionals such as emergency medicine and general medicine physicians who regularly review images in their daily practice.

Recruitment information / eligibility

• Status: Active, not recruiting
• Enrollment: 33
• Est. completion date: June 2025
• Est. primary completion date: October 2024
• Accepts healthy volunteers: Accepts Healthy Volunteers
• Gender: All
• Age group: N/A and older

Eligibility

Inclusion Criteria:

• General radiologists/radiographers/physicians who review CXRs as part of their routine clinical practice

Exclusion Criteria:

• Thoracic radiologists

• Non-radiology physicians with previous formal postgraduate CXR reporting training.

• Non-radiology physicians with previous career in radiology, respiratory medicine or thoracic surgery to registrar or consultant level

Related Conditions & MeSH terms

• Atelectasis
• Cardiomegaly
• Fibrosis Lung
• Multiple Pulmonary Nodules
• Pleural Effusion
• Pleural Effusions, Chronic
• Pneumoperitoneum
• Pneumothorax
• Pneumothorax; Acute
• Pulmonary Atelectasis
• Pulmonary Fibrosis
• Pulmonary Nodules, Multiple
• Solitary Pulmonary Nodule

Intervention

Other:
• Cases reading
The reading will be done remotely via the Report and Image Quality Control site (www.RAIQC.com), an online platform allowing medical imaging viewing and reporting. Participants can work from any location, but the work must be done from a computer with internet access. For avoidance of doubt, the work cannot be performed from a phone or tablet. The project is divided into two phases and participants are required to complete both phases. The estimated total involvement in the project is up to 20-24 hours. Phase 1: Time allowed: 2 weeks - Review 500 chest X-rays and express a clinical opinion through a structured reporting template (multiple choice, no open text required). Rest/washout period of 2 weeks. Phase 2 - Time allowed: 2 weeks - Review 500 chest X-rays together with an AI report for each case and express your clinical opinion through the same structured reporting template used in Phase A.
• Ground truthing
Two consultant thoracic radiologists will independently review the images to establish the 'ground truth' findings on the CXRs, where a consensus is reached this will then be used as the reference standard. In the case of disagreement, a third senior thoracic radiologist's opinion (>20 years experience) will undertake arbitration. A difficulty score will be assigned to each abnormality by the ground truthers using a 4-point Likert scale (1 being easy/obvious to 4 being hard/poorly visualised).

Locations

Country	Name	City	State
United Kingdom	Oxford University Hospitals NHS Foundation Trust	Oxford	Oxfordshire

Sponsors (1)

Lead Sponsor	Collaborator
Oxford University Hospitals NHS Trust

Country where clinical trial is conducted

United Kingdom, 

References & Publications (12)

Ahmad HK, Milne MR, Buchlak QD, Ektas N, Sanderson G, Chamtie H, Karunasena S, Chiang J, Holt X, Tang CHM, Seah JCY, Bottrell G, Esmaili N, Brotchie P, Jones C. Machine Learning Augmented Interpretation of Chest X-rays: A Systematic Review. Diagnostics (Basel). 2023 Feb 15;13(4):743. doi: 10.3390/diagnostics13040743. — View Citation

Greenhalgh R, Howlett DC, Drinkwater KJ. Royal College of Radiologists national audit evaluating the provision of imaging in the severely injured patient and compliance with national guidelines. Clin Radiol. 2020 Mar;75(3):224-231. doi: 10.1016/j.crad.2019.10.025. Epub 2019 Dec 19. — View Citation

Hillis JM, Bizzo BC, Mercaldo S, Chin JK, Newbury-Chaet I, Digumarthy SR, Gilman MD, Muse VV, Bottrell G, Seah JCY, Jones CM, Kalra MK, Dreyer KJ. Evaluation of an Artificial Intelligence Model for Detection of Pneumothorax and Tension Pneumothorax in Chest Radiographs. JAMA Netw Open. 2022 Dec 1;5(12):e2247172. doi: 10.1001/jamanetworkopen.2022.47172. — View Citation

Homayounieh F, Digumarthy S, Ebrahimian S, Rueckel J, Hoppe BF, Sabel BO, Conjeti S, Ridder K, Sistermanns M, Wang L, Preuhs A, Ghesu F, Mansoor A, Moghbel M, Botwin A, Singh R, Cartmell S, Patti J, Huemmer C, Fieselmann A, Joerger C, Mirshahzadeh N, Muse V, Kalra M. An Artificial Intelligence-Based Chest X-ray Model on Human Nodule Detection Accuracy From a Multicenter Study. JAMA Netw Open. 2021 Dec 1;4(12):e2141096. doi: 10.1001/jamanetworkopen.2021.41096. — View Citation

Jones CM, Buchlak QD, Oakden-Rayner L, Milne M, Seah J, Esmaili N, Hachey B. Chest radiographs and machine learning - Past, present and future. J Med Imaging Radiat Oncol. 2021 Aug;65(5):538-544. doi: 10.1111/1754-9485.13274. Epub 2021 Jun 25. — View Citation

Kundu R, Das R, Geem ZW, Han GT, Sarkar R. Pneumonia detection in chest X-ray images using an ensemble of deep learning models. PLoS One. 2021 Sep 7;16(9):e0256630. doi: 10.1371/journal.pone.0256630. eCollection 2021. — View Citation

Matsumoto T, Kodera S, Shinohara H, Ieki H, Yamaguchi T, Higashikuni Y, Kiyosue A, Ito K, Ando J, Takimoto E, Akazawa H, Morita H, Komuro I. Diagnosing Heart Failure from Chest X-Ray Images Using Deep Learning. Int Heart J. 2020 Jul 30;61(4):781-786. doi: 10.1536/ihj.19-714. Epub 2020 Jul 18. Erratum In: Int Heart J. 2020;61(5):1088. — View Citation

Spiritoso R, Padley S, Singh S. Chest X-ray interpretation in UK intensive care units: A survey 2014. J Intensive Care Soc. 2015 Nov;16(4):339-344. doi: 10.1177/1751143715580141. Epub 2015 May 18. — View Citation

van Beek EJR, Ahn JS, Kim MJ, Murchison JT. Validation study of machine-learning chest radiograph software in primary and emergency medicine. Clin Radiol. 2023 Jan;78(1):1-7. doi: 10.1016/j.crad.2022.08.129. Epub 2022 Sep 25. — View Citation

Wilson C. X-ray misinterpretation in urgent care: where does it occur, why does it occur, and does it matter? N Z Med J. 2022 Apr 1;135:49-65. — View Citation

Wu JT, Wong KCL, Gur Y, Ansari N, Karargyris A, Sharma A, Morris M, Saboury B, Ahmad H, Boyko O, Syed A, Jadhav A, Wang H, Pillai A, Kashyap S, Moradi M, Syeda-Mahmood T. Comparison of Chest Radiograph Interpretations by Artificial Intelligence Algorithm vs Radiology Residents. JAMA Netw Open. 2020 Oct 1;3(10):e2022779. doi: 10.1001/jamanetworkopen.2020.22779. — View Citation

Zech JR, Badgeley MA, Liu M, Costa AB, Titano JJ, Oermann EK. Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study. PLoS Med. 2018 Nov 6;15(11):e1002683. doi: 10.1371/journal.pmed.1002683. eCollection 2018 Nov. — View Citation

* Note: There are 12 references in all — Click here to view all references

Outcome

Type	Measure	Description	Time frame	Safety issue
Primary	Performance of AI algorithm: sensitivity	Evaluation of the Lunit INSIGHT CXR algorithm will be performed comparing it to the reference standard in order to determine sensitivity.	During 4 weeks of reading time
Primary	Performance of AI algorithm: specificity	Evaluation of the Lunit INSIGHT CXR algorithm will be performed comparing it to the reference standard in order to determine specificity.	During 4 weeks of reading time
Primary	Performance of AI algorithm: Area under the ROC Curve (AU ROC)	Evaluation of the Lunit INSIGHT CXR algorithm will be performed comparing it to the reference standard. Continuous probability score from the algorithm will be utilized for the ROC analyses, while binary classification results with a predefined operating cut-off will be used for evaluation of sensitivity, specificity, positive predictive value, and negative predictive value.	During 4 weeks of reading time
Primary	Performance of readers with and without AI assistance: Sensitivity	The study will include two sessions (with and without AI overlay), with all 30 readers reviewing all 500 CXR cases each time separated by a washout period to mitigate recall bias. The cases will be randomised between the two reads and for every reader.	During 4 weeks of reading time
Primary	Performance of readers with and without AI assistance: Specificity	The study will include two sessions (with and without AI overlay), with all 30 readers reviewing all 500 CXR cases each time separated by a washout period to mitigate recall bias. The cases will be randomised between the two reads and for every reader.	During 4 weeks of reading time
Primary	Performance of readers with and without AI assistance: Area under the ROC Curve (AU ROC)	The study will include two sessions (with and without AI overlay), with all 30 readers reviewing all 500 CXR cases each time separated by a washout period to mitigate recall bias. The cases will be randomised between the two reads and for every reader.	During 4 weeks of reading time
Primary	Reader speed with vs without AI assistance.	Mean time taken to review a scan, with vs without AI assistance.	During 4 weeks of reading time

External Links

•   Richards M. Diagnostics: Recovery and Renewal - Report of the Independent Review of Diagnostic Services for NHS England. NHS England 2022.
•   Royal College of Radiologists. Clinical radiology UK workforce census 2019 report. Royal College of Radiologists 2020.
•   www.nice.org.uk. (2022). Summary | Artificial intelligence for analysing chest X-ray images | Advice | NICE. [online] Available at: