Related Experiment Video
Updated: Jun 16, 2026
08:09
Transthoracic Echocardiography in Mice
Published on: May 28, 2010
62.1K
Automated echocardiographic left ventricular dimension assessment in dogs using artificial intelligence: Development
Catherine C Stowell1, Valeria Kallassy2, Beth Lane1
1National Heart and Lung Institute (NHLI), Imperial College, London, UK.
Journal of Veterinary Internal Medicine
|February 16, 2024
Summary
Artificial intelligence (AI) can accurately measure canine echocardiographic left ventricular (LV) dimensions. The AI’s performance was comparable to veterinary experts, showing potential for improved accuracy and reproducibility in canine cardiology.
Area of Science:
- Veterinary cardiology
- Artificial intelligence in medicine
- Medical imaging analysis
Background:
- Echocardiographic measurements in dogs can lack accuracy and reproducibility.
- Artificial intelligence (AI) offers potential to enhance these measurements.
Purpose of the Study:
- To train a neural network for measuring left ventricular (LV) linear dimensions in canine echocardiograms.
- To evaluate the AI's performance against expert measurements.
Main Methods:
- A neural network was trained on 1398 echocardiogram frames from 461 dogs.
- AI measurements were compared to consensus expert annotations on 50 validation studies.
- Accuracy was quantified by deviation from expert consensus.
Main Results:
- AI measurements showed no significant difference from expert consensus for septal dimensions in end-diastole.
- AI performance was comparable or superior to experts for LV dimensions (LVD) and free wall measurements.
- No performance differences were observed between AI and experts in end-systole.
Conclusions:
- An AI network can be effectively trained to measure linear LV dimensions in dogs.
- The AI's measurement performance is indistinguishable from that of experienced veterinary cardiologists.
- AI holds promise for improving accuracy and reproducibility in canine echocardiography.
