loading page

An attention-based deep learning method for right ventricular quantification using 2D echocardiography: feasibility and accuracy
  • +8
  • Polydoros Kampaktsis,
  • Tuan A. Bohoran,
  • Laura McLaughlin,
  • Jay Leb S,
  • Zhonghua Liu,
  • Serafeim Moustakidis,
  • Athanasios Siouras,
  • Anvesha Singh,
  • Rebecca Hahn,
  • Gerry P. McCann,
  • Archontis Giannakidis
Polydoros Kampaktsis
Columbia University Irving Medical Center

Corresponding Author:pkampaktsis@yahoo.com

Author Profile
Tuan A. Bohoran
Nottingham Trent University
Author Profile
Laura McLaughlin
Columbia University Irving Medical Center
Author Profile
Jay Leb S
Columbia University Department of Radiology
Author Profile
Zhonghua Liu
Columbia University Department of Biostatistics
Author Profile
Serafeim Moustakidis
Panepistemio Thessalias Tmema Iatrikes
Author Profile
Athanasios Siouras
Panepistemio Thessalias Tmema Iatrikes
Author Profile
Anvesha Singh
University of Leicester Department of Cardiovascular Sciences
Author Profile
Rebecca Hahn
Columbia University Division of Cardiology
Author Profile
Gerry P. McCann
University of Leicester Department of Cardiovascular Sciences
Author Profile
Archontis Giannakidis
Nottingham Trent University
Author Profile

Abstract

Aim: To test the feasibility and accuracy of a new attention-based deep learning (DL) method for right ventricular (RV) quantification using 2D echocardiography (2DE) with cardiac magnetic resonance imaging (CMR) as reference. Methods and results: We retrospectively analyzed images from 50 adult patients (median age 51, interquartile range 32-62 42% women) who had undergone CMR within 1 month of 2DE. RV planimetry of the myocardial border was performed in end-diastole (ED) and end-systole (ES) for 8 standardized 2DE RV views with calculation of areas. The DL model comprised a Feature Tokenizer module and a stack of Transformer layers. Age, gender and calculated areas were used as inputs, and the output was RV volume in ED/ES. The dataset was randomly split into training, validation and testing subsets (35, 5 and 10 patients respectively). Mean RVEDV, RVESV and RV ejection fraction (EF) were 163±70ml, 82±42ml and 51±8% respectively without differences among the subsets. The proposed method achieved good prediction of RV volumes (R 2=0.953, absolute percentage error [APE]=9.75±6.23%) and RVEF (APE=7.24±4.55%). Per CMR, there was 1 patient with RV dilatation and 3 with RV dysfunction in the testing dataset. The DL model detected RV dilatation in 1/1 case and RV dysfunction in 4/3 cases. Conclusions: An attention-based DL method for 2DE RV quantification showed feasibility and promising accuracy. The method requires validation in larger cohorts with wider range of RV size and function. Further research will focus on the reduction of the number of required 2DE to make the method clinically applicable.
09 Aug 2023Submitted to Echocardiography
09 Aug 2023Submission Checks Completed
09 Aug 2023Assigned to Editor
24 Aug 2023Reviewer(s) Assigned
10 Sep 2023Review(s) Completed, Editorial Evaluation Pending
10 Sep 2023Editorial Decision: Revise Minor
30 Oct 20231st Revision Received
01 Nov 2023Submission Checks Completed
01 Nov 2023Assigned to Editor
01 Nov 2023Reviewer(s) Assigned
01 Nov 2023Review(s) Completed, Editorial Evaluation Pending
05 Nov 2023Editorial Decision: Accept