Semester

Spring

Date of Graduation

2025

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Not Listed

Committee Chair

Donald Adjeroh

Committee Co-Chair

Gianfranco Doretto

Committee Member

Gianfranco Doretto

Committee Member

Prashnna Gyawali

Abstract

Left ventricular ejection fraction (LVEF) is a critical biomarker for heart failure, but manual estimation from echocardiograms is time-consuming. Artificial intelligence can be used to accelerate this process, allowing clinicians to focus on other critical tasks. Current methods typically train models from scratch on echocardiogram datasets; however, this approach is limited by the scarcity of large medical imaging datasets, which are expensive and difficult to acquire. We present a transfer learning approach that leverages pretrained models from massive datasets, enabling continuous improvement as foundation models advance. Our method employs visual prompting to generate trainable masks for echocardiogram videos, transforming the input data into representations compatible with the backbone model’s classification space. When evaluated on the EchoNet Dynamic dataset using CLIP as the backbone model, our framework achieves an AUROC of 0.89. This lightweight solution provides reliable LVEF prediction while avoiding expensive retraining, demonstrating the potential of adapting general-purpose vision models for specialized medical applications through innovative prompting techniques.

Download

Share

COinS