A5101719622- Fetal and Pediatric Neurological Disorders
- Domain Adaptation and Few-Shot Learning
- Cardiovascular Function and Risk Factors
- Cardiac Imaging and Diagnostics
- Advanced MRI Techniques and Applications
- Medical Image Segmentation Techniques
- Advanced Neural Network Applications
- Coronary Interventions and Diagnostics
- Radiomics and Machine Learning in Medical Imaging
- Cleft Lip and Palate Research
- Medical Imaging Techniques and Applications
- Prenatal Screening and Diagnostics
- AI in cancer detection
- Artificial Intelligence in Healthcare and Education
- Cardiac Valve Diseases and Treatments
- Medical Imaging and Analysis
- Generative Adversarial Networks and Image Synthesis
- Cardiovascular Disease and Adiposity
- Autopsy Techniques and Outcomes
- COVID-19 diagnosis using AI
- Pressure Ulcer Prevention and Management
- Congenital Heart Disease Studies
- Elasticity and Material Modeling
- Neonatal and fetal brain pathology
- Geriatric Care and Nursing Homes
Imperial College London
2017-2024
HeartFlow (United States)
2020-2024
Institute of Group Analysis
2017-2022
King's College London
2015-2021
Imperial Valley College
2018
London Women's Clinic
2018
University of Illinois Urbana-Champaign
2016
North Shore University Hospital
2016
British Heart Foundation
2015
Engineering and Physical Sciences Research Council
2015
Cardiovascular resonance (CMR) imaging is a standard modality for assessing cardiovascular diseases (CVDs), the leading cause of death globally. CMR enables accurate quantification cardiac chamber volume, ejection fraction and myocardial mass, providing information diagnosis monitoring CVDs. However, years, clinicians have been relying on manual approaches image analysis, which time consuming prone to subjective errors. It major clinical challenge automatically derive quantitative clinically...
This study sought to develop a fully automated framework for cardiac function analysis from magnetic resonance (CMR), including comprehensive quality control (QC) algorithms detect erroneous output.Analysis of cine CMR imaging using deep learning (DL) could automate ventricular assessment. However, variable image quality, variability in phenotypes disease, and unavoidable weaknesses training DL currently prevent their use clinical practice.The consists pre-analysis QC, followed by algorithm...
Patient-specific models of blood flow in the coronary arteries have entered clinical practice worldwide to aid diagnosis and management heart disease. This technology leverages modern AI-based image segmentation methods extract geometry from noninvasive computed tomography volumetric imaging, computational physiology define boundary conditions fluid dynamic techniques compute artery pressure. The described herein been used fractional reserve (FFR) arteries, namely ratio pressure a reference...
Deep learning approaches such as convolutional neural nets have consistently outperformed previous methods on challenging tasks dense, semantic segmentation. However, the various proposed networks perform differently, with behaviour largely influenced by architectural choices and training settings. This paper explores Ensembles of Multiple Models Architectures (EMMA) for robust performance through aggregation predictions from a wide range methods. The approach reduces influence...
Measurement of head biometrics from fetal ultrasonography images is key importance in monitoring the healthy development fetuses. However, accurate measurement relevant anatomical structures subject to large inter-observer variability clinic. To address this issue, an automated method utilizing Fully Convolutional Networks (FCN) proposed determine measurements circumference (HC) and biparietal diameter (BPD). An FCN was trained on approximately 2000 2D ultrasound with annotations provided by...
Deep learning models for semantic segmentation are able to learn powerful representations pixel-wise predictions, but sensitive noise at test time and may lead implausible topologies. Image registration on the other hand warp known topologies target images as a means of segmentation, typically require large amounts training data, have not widely been benchmarked against models. We propose Atlas Image-and-Spatial Transformer Network (Atlas-ISTN), framework that jointly learns 2D 3D image...
Detecting acoustic shadows in ultrasound images is important many clinical and engineering applications. Real-time feedback of can guide sonographers to a standardized diagnostic viewing plane with minimal artifacts provide additional information for other automatic image analysis algorithms. However, automatically detecting shadow regions using learning-based algorithms challenging because pixel-wise ground truth annotation subjective time consuming. In this paper, we propose weakly...
Patient-specific models of blood flow are being used clinically to diagnose and plan treatment for coronary artery disease. A remaining challenge is bridging scales from in arteries the micro-circulation supplying myocardium. Previously proposed descriptive rather than predictive have not been applied human data. The goal here develop a multiscale patient-specific model enabling simulation large myocardial tissue. Patient vasculatures segmented computed tomography angiography data extended...
Tissue characterisation with CMR parametric mapping has the potential to detect and quantify both focal diffuse alterations in myocardial structure not assessable by late gadolinium enhancement. Native T1 particular shown promise as a useful biomarker support diagnostic, therapeutic prognostic decision-making ischaemic non-ischaemic cardiomyopathies. Convolutional neural networks Bayesian inference are category of artificial which model uncertainty network output. This study presents an...
We present a framework for combining cardiac motion atlas with non-motion data. The represents cycle across number of subjects in common space based on rich descriptors capturing 3D displacement, velocity, strain and rate. data are derived from variety sources such as imaging, electrocardiogram (ECG) clinical reports. Once the space, we apply novel supervised learning approach random projections ensemble to learn relationship between some desired output. our problem predicting response...
Advances in artificial intelligence (AI) have demonstrated potential to improve medical diagnosis. We piloted the end-to-end automation of mid-trimester screening ultrasound scan using AI-enabled tools.A prospective method comparison study was conducted. Participants had both standard and AI-assisted US scans performed. The AI tools automated image acquisition, biometric measurement, report production. A feedback survey captured sonographers' perceptions scanning.Twenty-three subjects were...
2D ultrasound (US) is still the preferred imaging method for fetal screening. However, biometrics are significantly affected by inter/intra-observer variability and operator dependence of a traditionally manual procedure. 3DUS an alternative emerging modality with potential to alleviate many these problems. This paper presents new automatic framework skull segmentation in 3DUS. We propose two-stage convolutional neural network (CNN) able incorporate additional contextual structural...
Cardiac motion atlases provide a space of reference in which the motions cohort subjects can be directly compared. Motion used to learn descriptors that are linked different pathologies and subsequently for diagnosis. To date, all such have been formed applied using data from same modality. In this work we propose framework build multimodal cardiac atlas 3D magnetic resonance (MR) ultrasound (US) data. Such an will benefit complementary features derived two modalities, furthermore, it could...
Cardiovascular magnetic resonance myocardial feature tracking (CMR-FT) is a promising method for quantification of cardiac function from standard steady-state free precession (SSFP) images. However, currently available techniques require operator dependent and time-consuming manual intervention, limiting reproducibility clinical use. In this paper, we propose fully automated pipeline to compute left ventricular (LV) longitudinal radial strain 2- 4-chamber cine acquisitions, LV...