Open Collections

Weakly supervised landmark detection for automatic measurement of left ventricular diameter in videos of PLAX from cardiac ultrasound

University of British Columbia

The length of the left ventricle and how it changes throughout a cardiac cycle is an indicative of the health of the heart in clinical studies. Currently, this measurement is computed through manual labeling made by experienced sonographers. This labeling procedure is highly costly, time-consuming, and prone to inter-observer and intra-observer errors. In order to overcome these challenges, this thesis proposes a deep neural network-based model to automate the detection of landmarks corresponding to the diameter of the heart. The dataset used for this work is sparse in the temporal dimension, having annotations merely available on end-diastolic and end-systolic frames, and the final goal of the model is to provide meaningful measurements across the entire cardiac cycle. The proposed network leads to 10.08% average ejection fraction error, and 12.18% mean percentile error, which is satisfactory based on the requirements of this project.

Text

2021-03-23

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/77582

Electrical and Computer Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/