A5008865726- Cardiovascular Function and Risk Factors
- Cardiac Imaging and Diagnostics
- Cardiac Valve Diseases and Treatments
- Cardiovascular Health and Disease Prevention
- Hemodynamic Monitoring and Therapy
- Ultrasound in Clinical Applications
- Congenital Heart Disease Studies
- Industrial Vision Systems and Defect Detection
- Aortic aneurysm repair treatments
- Pulmonary Hypertension Research and Treatments
- Artificial Intelligence in Healthcare and Education
- Aortic Disease and Treatment Approaches
- Image and Signal Denoising Methods
- Infective Endocarditis Diagnosis and Management
- Infrared Target Detection Methodologies
Imperial College London
2021-2025
Lung Institute
2025
Imperial College Healthcare NHS Trust
2020-2024
Hammersmith Hospital
2020-2021
Abstract Background/Introduction Improving uptake and accessibility of echocardiography necessitates more efficient training. Current training programmes are focused around expert mentorship, clinical practicals, lectures. All these resource-intensive, difficult to scale. Research into the way humans learn has consistently found that techniques such as spaced testing interleaving subjects, deliver superior retention - even complex topics. These can be most easily cost-effectively delivered...
Abstract Background Artificial intelligence (AI) measurement has the potential to transform cardiac imaging, but what if it makes a mistake? Can AI also highlight when is most likely have mismeasured and can experts improve these measurements? As readers become increasingly reliant on automated measures for echocardiographic analysis, we need confidence that software will measure accurately, flag where human oversight needed. Purpose To develop test an open, scientific, machine-learning...
Abstract Background A major challenge in real-world echocardiography is the difficulty obtaining high quality images some patients or clinical settings. Is AI only useful when image good? Purpose To artificially degrade adequate-quality images, and compare ability of human experts AI, to make measurements correctly, as degradation worsens. Methods PLAX dimension were made on videos 30 with a range LV dimensions (mean 138mm, SD 37mm). set gold standard, 9 measured each image, blinded by...
Background: Artificial intelligence (AI) for echocardiography requires training and validation to standards expected of humans. We developed an online platform established the Unity Collaborative build a dataset expertise from 17 hospitals training, validation, standardization such techniques. Methods: The consisted 2056 individual frames drawn at random 1265 parasternal long-axis video-loops patients undergoing clinical in 2015 2016. Nine experts labeled these images using our platform....
Global longitudinal strain (GLS) is reported to be more reproducible and prognostic than ejection fraction. Automated, transparent methods may increase trust uptake. The authors developed open machine-learning–based GLS methodology validate it using multiexpert consensus from the Unity UK Echocardiography AI Collaborative. We trained a multi-image neural network (Unity-GLS) identify annulus, apex, endocardial curve on 6,819 apical 4-, 2-, 3-chamber images. external validation dataset...
Abstract Background Artificial intelligence (AI) could improve accuracy and reproducibility of echocardiographic measurements in dogs. Hypothesis A neural network can be trained to measure left ventricular (LV) linear dimensions Animals Training dataset: 1398 frames from 461 canine echocardiograms a single specialist center. Validation: 50 additional the same Methods right parasternal 4‐chamber long axis frame each study, labeled by 1 18 echocardiographers, marking anterior posterior points...
Doppler echocardiography is a widely utilised non-invasive imaging modality for assessing the functionality of heart valves, including mitral valve. Manual assessments traces by clinicians introduce variability, prompting need automated solutions. This study introduces an innovative deep learning model detection peak velocity measurements from inflow images, independent Electrocardiogram information. A dataset images annotated multiple expert cardiologists was established, serving as robust...
Abstract Background The longitudinal function of the left ventricle can be quantified by measuring myocardial excursion with or without normalising diastolic length to derive strain. Furthermore, these lengths determined from curved in a straight-line mid-annulus apex parallel long axis. All four combinations have been advocated and used, but optimal is unknown. Purpose Compare prognostic information contained measures large retrospective cohort, measurements performed AI. Methods cohort...
Abstract Background Surveillance of aortic dimensions requires reproducible measurements, and knowledge the rate progression in those with dilatation. Purpose To develop an open machine-learning method for measuring root proximal ascending aorta on echocardiograms, validate it through a multi-expert panel from Unity UK Echocardiography AI Collaborative, applying to derive historical databases. Methods The neural network was trained 1478 parasternal long-axis images. Each image labelled key...
Objectives: Tricuspid annuloplasty is the optimal surgical repair technique for tricuspid regurgitation which improves mortality and morbidity. Ring annuloplasties techniques of choice. Here, we evaluate efficacy durability a new method interrupted pledgeted suture annuloplasty. Methods: Between 2011 2018, 39 eligible patients underwent valve using this novel technique. Indication was grade at moderate or greater, an annular diameter >40 mm. Patients were assessed both preoperatively...
Abstract Background and purpose Artificial intelligence (AI) has the potential to greatly improve efficiency reproducibility of quantification in echocardiography, but gain widespread use it must both meet expert standards excellence have a transparent methodology. We developed an online platform enable multiple collaborators annotate medical images for training validating neural networks. Methods Using our collaborative 9 echocardiographers labelled 2056 that comprised dataset. They four...
Abstract Background Left ventricular longitudinal strain has been reported to deliver reproducibility, sensitivity and prognostic value over above ejection fraction. However, it currently relies on uninspectable proprietary algorithms suffers from a lack of widespread clinical use. Uptake may be improved by increasing user trust through greater transparency. Purpose We therefore developed machine-learning based method, trained, validated with accredited experts our AI Echocardiography...