Previously submitted to: JMIR Research Protocols (no longer under consideration since Jun 30, 2026)
Date Submitted: Jun 17, 2025
Open Peer Review Period: Jun 17, 2025 - Aug 12, 2025
(closed for review but you can still tweet)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Identify Right Ventricular Dysfunction from Chest X- Ray Using Deep Learning:Protocol for a Feasibility Trial
ABSTRACT
Background:
Right ventricular dysfunction (RVD) is an important predictor of outcomes in patients with heart disease, which has significant prognostic implications for a range of diseases, including ischaemic and non-ischaemic cardiomyopathy, valvular heart disease, congenital heart disease, and pulmonary hypertension. The routine diagnosis of right heart structural abnormalities at an earlier stage, ideally when only structural abnormalities are present but patients are not yet symptomatic, represents an elusive but critical goal in the field of cardiology.
Objective:
The objective of this study is to apply deep learning (DL) analysis to chest X-rays (CXRs) in order to accurately detect specific structural abnormalities, thus facilitating the early identification of patients exhibiting RVD and improving outcomes.
Methods:
Approximately 10,000 adult CXRs with corresponding echocardiographic labels of right ventricular function, including right ventricular fractional area change (RVFAC), tricuspid annular plane systolic excursion (TAPSE) and right ventricular myocardial performance index (RVMPI), are currently being collected within a 12-month period. The study will employ DenseNet-121 for the interpretation of CXRs with the objective of identifying the presence of RVD. The DL model will be evaluated against independently collected and labelled sets of CXRs. Subsequently, three main assessments of the model will be performed: validation of the trained model on a separate dataset of patients who have been treated in The emergency department, validation on another independent test dataset obtained from healthy volunteers, and comparison of the model's performance against that of five radiologists on a sample of CXRs will be conducted. The primary objective of the study protocol is the creation of a DL model that is capable of accurately identifying RVD from inexpensive and prevalent examinations of CXRs.
Results:
The collected data will be synthesized to evaluate the program's acceptability and the feasibility of study procedures. Ethical approval was obtained in November 2024, followed by the initiation of participant recruitment. Data collection is scheduled to continue through December 2025.
Conclusions:
This feasibility trial may confirm deep learning can identify right ventricular dysfunction from chest X-rays Clinical Trial: Trial registration number Chinese Clinical Trial Registry, ChiCTR2500095838.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.