Echocardiography generation using deep generative models
Abstract
Echocardiography is a widely used, low-cost, and non-invasive imaging modality for the assessment and management of cardiac disease. It provides real-time visualisation of cardiac anatomy and function and supports the estimation of clinically important physiological parameters such as ejection fraction. Despite its clinical utility, echocardiography is highly operator-dependent, particularly for transesophageal echocardiography (TEE), and the availability of large, well-annotated datasets remains limited due to acquisition complexity, data protection constraints, and the cost of expert annotation. These challenges hinder the development and evaluation of datadriven methods for echocardiographic analysis. This thesis investigates the use of deep generative models to address data scarcity and standardisation challenges in echocardiography. First, a pipeline termed HeartTo Echo is proposed to generate anatomically consistent pseudo-images corresponding to 21 standard TEE views from three-dimensional heart models. These pseudoimages can be translated into realistic synthetic echocardiograms, as demonstrated for the mid-esophageal four-chamber view, and their utility is evaluated through qualitative assessment and improved performance on a downstream left-ventricle segmentation task. Building on this, a data-efficient diffusion-based framework, EchoLoRA, is introduced for synthetic TEE image generation from semantic masks. By combining low-rank adaptation with pretrained…
Read the paper
Related studies
- Learning from First-Generation Qualitative Approaches in the IS Discipline: An Evolutionary View and Some Implications for Authors and Evaluators (PART 1/2)
- Commentary on the clinical and preclinical dosage limits of interstitially administered magnetic fluids for therapeutic hyperthermia based on current practice and efficacy models
- Prosthesis: New Technological Opportunities and Innovative Biomedical Devices
- The history and future of neural modeling for cochlear implants
- Ryanodine receptors: waking up from refractoriness
- Novel Transgenic Rabbit Model Sheds Light on the Puzzling Role of Matrix Metalloproteinase-12 in Atherosclerosis
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.